A comitative source for object markers in Sinitic languages: 跟 kai⁵⁵ in Waxiang and 共 kang⁷ in Southern Min

Abstract

This analysis sets out to specifically discuss the polyfunctionality of 跟 [kai⁵⁵] in Waxiang (Sinitic), whose lexical source is the verb ‘to follow’. Amongst its various uses, we find a preposition ‘with, along’, a marker of adjuncts and a NP conjunction, thus superficially resembling its Mandarin cognate 跟 gēn ‘with’. Curiously, however, it has also evolved into a direct object marker in Waxiang, with a function similar to that of preposition 把 bă < ‘hold, take’ as found in the S–bă–O–VP or so-called ‘disposal’ form in standard Mandarin. The pathways of grammaticalization for 跟 [kai⁵⁵] in Waxiang are thus discussed in order to determine how it has developed this unusual grammatical function in one of the linguistic zones of China where verbs of giving or taking are, in fact, the main source for grammaticalized object markers in ‘disposal’ constructions.

On the basis of 16^th and 17^th century Southern Min literature (Sinitic), a comparison is also made with analogous developments for comitative 共 gòng ‘with’ to provide support for our hypothesis that the direct object marking use has evolved from the oblique function of a benefactive or dative, and is clearly separate from the crosslinguistically well-attested pathway that leads to its use as a conjunction. We would thus like to propose that these data contribute a new pattern to the stock of grammaticalization pathways, specifically, comitative > dative/benefactive > accusative (direct object marker).

In this analysis, we propose to investigate the intriguing case of comitative prepositions that come to be used as direct object markers, particularly with respect to 跟 kai⁵⁵ ‘with’ in Waxiang, an unclassified Sinitic language spoken in Western Hunan, China. Our aim is to make certain conjectures as to the pathway for this diachronic change, considering both morphosyntactic and semantic features from within the framework of context-induced grammaticalization (Heine 2002). With this purpose in mind, we also make reference to 16^th-17^th century manuscripts from Southern Min, another Sinitic language spoken largely in Fujian and in Taiwan, in order to trace similar developments for its comitative preposition 共 gòng, a comitative with a different source from 跟 kai⁵⁵.

Prepositional constructions derived from serial verb constructions in Chinese, especially the object-marking types, have attracted a great deal of attention in Chinese linguistics over the past few decades. Peyraube (1989, 1991), Mei Tsu-lin (1990), Wei Pei-chuan (1997), Cao Guangshun & Yu Hsiao-jung (2000), Liu Ziyu (2002), Wu Fuxiang (2003a/2006, 2003b), Cao Guangshun & Long Guofu (2005), and Jiang Shaoyu (2008), have all carefully described the pathways of grammaticalization for object markers on the basis of textual material, particularly for the so-called ‘disposal marker’ 把 bă in the Archaic and Medieval Chinese periods.

The ‘disposal form’ (chŭzhìshì) is best known in the form of the 把 bă construction of Standard Mandarin Chinese and has the syntactic configuration: NP_1(Agent) – Object Marker –NP_2(Patient)–VP, where the object marker is a preposition. While this marker is typically 把 bă in most of the Northern Sinitic area, an array of forms is found in the other main dialect groups belonging to Sinitic. It creates a strong contrast to one of the basic word orders in Chinese languages of S–V–O.

Chappell (2000, 2006, 2007) has carried out research on this topic for contemporary Sinitic languages in order to examine the diversity of lexical sources as well as the evolution of these object markers. According to her study, there are three main sources for object markers in Sinitic languages. These are the following (Chappell 2006: 469):

1. Verbs of taking and holding > object markers, e.g. cognates and synonyms of 把 bă, 将 jiāng and 拿 ná ‘to take’ as in Standard Mandarin and Jin dialects

2. Verbs of giving and helping > object markers, e.g. cognates and synonyms of 给 gĕi ‘to give’ and 帮 bāng ‘to help’, as found in many Wu, Hui, Xiang and Southwestern Mandarin dialects

3. Comitatives > object markers, e.g. cognates and synonyms of 共 gòng in the Min dialects, 㧯 lau² or 同 tung² in Hakka; and also comitatives in the Wu dialects, according to Xu Baohua & Tao Huan (1998) and Hu Songbo & Ge Xin (2003), and in a few non-standard Mandarin dialects and in a small number of Gan dialects.

Chappell also points out that while the detailed stages of the grammaticalization pathways have been well described in the case of verbs of taking and holding in Chinese, they are still to be worked out in detail for the two additional sources of give/help verbs and comitatives. In addition to this, from a typological point of view, comitatives used as object markers are rare cross-linguistically, and rare in Sinitic languages too.¹ Note that several sources for these comitatives are implicated.

The lexeme 跟 kai⁵⁵ in the Waxiang dialect can be used as a verb ‘to follow’, a comitative preposition ‘with’, a NP conjunction ‘and’, a benefactive ‘for’ and an object marker, among other uses. Unusually, it has developed this object-marking use in an area of China where either the marker 把 bă or verbs of giving and helping are in fact the main source for the object markers, as found in the local varieties of Southwestern Mandarin spoken in western Hunan.² This analysis thus sets out to determine the precise pathway of grammaticalization for 跟 kai⁵⁵ in the Waxiang dialect, hypothesizing, through comparison with Southern Min, that this has taken place by means of extension of meaning from the benefactive/dative use. To this purpose, historical data on Southern Min from 16^th – 17^th centuries will be used, non-existent for Waxiang.

The layout of the presentation is as follows: background information on the Waxiang language is provided in §1 while §2 and §3 present data from two varieties of Waxiang –Guzhang and Yuanling; §4 discusses all the lexical and grammaticalized functions of Waxiang 跟 kai⁵⁵, while §5 contains a description of object markers and comitatives in Hunan and other central and southeastern provinces of China. In §6, the Southern Min historical materials are examined for the usage of comitative 共 gòng and finally in §7, the proposed grammaticalization chains are discussed.

1 Languages of Guzhang County, Western Hunan

Waxiang is a dying language spoken in a remote mountainous area of western Hunan, known as Xiangxi (湘西), designated as a bilingual area in 1985 (Bao Houxing & Li Yongming, 1985).³ Like most Sinitic languages, it is tonal, analytic in tendency and has S-V-O as a basic word order (Wu & Shen 2010). It has developed in a contact situation surrounded by genetically unrelated languages, in particular, Hmong (Hmong-Mien) and Tujia (Tibeto-Burman), while retaining an archaic core of Chinese features. Although the heartland of the Xiang group of dialects is located in Hunan, there has been little contact between Waxiang and Xiang speakers over the centuries, the two groups inhabiting non-contiguous areas. In fact, for Waxiang speakers today, fluency in a second language will typically be in Southwestern Mandarin.

Three main dialects of Waxiang have been investigated: these are Yuanling 沅 陵, Guzhang 古丈 and Luxi 泸溪 (for phonology only in the latter case). Waxiang speakers are in fact located across five counties in Xiangxi Tujia and Miao Autonomous Region, for which a detailed description is given in Wu & Shen (2010:1-5).

Archaic features of Old Chinese preserved by Waxiang not only pertain to its sound system (Y. Wu 2000), but also to lexical items including the terms for ‘shoe’ which is 履 li²⁵, ‘wok’ which is 鼎 taŋ¹³, ‘basket’ which is 豆 da³³and the verb ‘to like, to love’ 字 dza³³. Other distinguishing features are its tripartite demonstrative paradigm and the use of tone sandhi to express the plural form of the personal pronouns. The aspect system is in the primary stage such that no completely grammaticalized and semantically bleached perfective markers are to be found (Wu & Shen, 2010).

The sociolinguistic picture in the Waxiang-speaking areas is very complex. This can be readily perceived in an important historical record, the Guzhangping Tingzhi (Guzhangping Gazetteer) compiled in 1907 by Dong Hongxun 董鸿勋, which contains a detailed description of the languages and dialects spoken in each village in Guzhang county, western Hunan, at this time. He classified the local population into five categories according to their origins:

1. Hmong (or Miao) 苗

2. Gelao 仡佬⁴

3. Tujia 土家

4. Min, 民 which means ‘the people’ or those who speak Waxiang and

5. Ke, 客 which means ‘guests’ or those who speak the Southwestern variety of Mandarin.

A map of Hunan is presented below to show the overall distribution of Sinitic, Tibeto-Burman and Hmong languages as spoken in this province, including this western area of Xiangxi, the target of our research.

Some interesting trends which emerge from the Gazetteer are that no-one in the Mandarin community was able to speak Waxiang in 1907. Furthermore, while a mere 13% of Waxiang people had shifted to speaking only Mandarin in 1907, even less, just 9%, were bilingual. In other words, at the end of the 19^th century, 78% of Waxiang speakers exclusively spoke their own language. In general, local non-Han peoples adopted Southwestern Mandarin earlier than did the Waxiang speakers, given that 72% of people with minority origins, mainly the Hmong and the Tujia, could speak Mandarin as their second language as early as 1907.

Map 2 below represents the language situation of Guzhang in 1907 according to Dong Hongxun’s description. This should be compared with Map 3 which represents the language situation for Guzhang in 1998 based on Yunji Wu’s study (Y. Wu, 2000: 352–353). Map 3 clearly shows that nearly all present-day Waxiang speakers can now speak Mandarin, a sign of language shift in progress.

2 The usages of 跟 kai⁵⁵ in the Guzhang Waxiang dialect

We carried out a detailed investigation of the uses of 跟 gēn, realised as kai⁵⁵ in the Guzhang Waxiang dialect, and discovered many examples in both narratives and conversations, collected principally during fieldwork trips in 2005, 2006, 2008 and 2010. The lexeme 跟 gēn has the following eight functions in the Guzhang Waxiang dialect. The first function is as a verb ‘to follow’ as found in Mandarin and many other Sinitic languages. In this use, it may be modified by aspectual and directional complements, as shown in (1).⁵

2.1. Lexical verb meaning ‘to follow’:

Its second function as a comitative preposition is also shared with Mandarin 跟 gēn:

2.2. Comitative preposition ‘with’:

2.3. Coordinative conjunction ‘and’

In the Waxiang dialects, 跟 kai⁵⁵ can be used as a coordinative conjunction, as its cognate 跟 gēn can in standard Mandarin:

Note however that this is not the only coordinative conjunction, as 跟 kai⁵⁵ in this function can be found used alongside another marker 邀 iau⁵⁵ ‘and’.

2.4. Benefactive marker meaning ‘(to do something) for someone’.

This use is similar to one of the grammaticalized functions of Mandarin 给 gĕi ‘give’:

2.5. Dative marker

As a dative marker, 跟 kai⁵⁵ introduces the indirect object with verbs of giving or the addressee with verbs of communication; each verb class is exemplified below:

2.6. Directional preposition

跟 kai⁵⁵ can also be used with a following locative NP to introduce the direction of an action, translatable as ‘towards’ or ‘along’. In this function, it corresponds to Mandarin 向 xiàng ‘towards’ and 顺 shùn ‘along’ or yán 沿 ‘along’:

2.7. Source locative preposition ‘from’

跟 kai⁵⁵ can also introduce a reference to location that is the source when preceding a place noun, ‘from’. It thus resembles the uses of Mandarin 从 cóng ‘from’:

2.8. Object marker

The lexeme 跟 kai⁵⁵ can also be used in the same way as Mandarin 把 bă to introduce the direct object, forming a highly transitive S- kai⁵⁵-O-VP construction. Thus, a common and important function of 跟 kai⁵⁵ in Waxiang is as an object marker in the construction [gen_om + NP + VP]. In fact, this is its most frequent use, as Table 1 shows, based on transcriptions of eight narratives by Mr Xiang Guangxun (20:27 minutes in total), recorded in Guzhang during fieldwork.

Table 1.

Frequency of kai⁵⁵ in Guzhang Waxiang oral texts¹

Function	Verb	Comitative	Benefactive	Accusative object	Total
Frequency	2	2	6	33	43

¹These were the only 4 uses found in the particular texts. That the other 4 were not present is an accidental gap in the data.

Given that this is the principal construction with 跟 kai⁵⁵ we wish to treat, an extended set of examples is presented below. In fact, the object construction with 跟 kai⁵⁵ may co-occur with a large array of different predicates, including quite elaborate ones with complex VPs such as those containing postverbal nouns. Similarly to Mandarin 把 bă, the 跟 gēn object-marking construction typically codes a telic predicate by means of perfective or bounded kinds of aspect marking, if not by resultative and directional verb complexes. Telic predicates express the completion of changes of state, including changes in location, as the following examples demonstrate.

A resultative verb compound, ‘wound by biting’, can be found in example (10) accompanied by the perfective aspect marker ti:

The predicate in the following example, (11), contains a monosyllabic verb, to⁵⁵ ‘take’, followed by the directional complement ts literally ‘out-come’. The whole forms a directional verb complex, expressing a change in location for the direct object, the quilt.

Even in imperatives, monosyllabic verbs usually occur with either a deictic motion verb such as ‘come’ which codes motion towards the speaker; or with an aspect marker, ka³³ for semelfactive action, as in (12). That is, bare verbs are rarely found in our texts, similar to the case for standard Mandarin 把 bă.

In the four following examples, a postverbal element is present, representing the so-called ‘give’, ‘place’ and ‘make’ types of ‘disposal’ construction identified in Medieval Chinese by Chen Chusheng (1983), Ye Youwen (1988) and Mei Tsu-lin (1990) as three of the early types of object-marking constructions. Thus, examples (13) and (14) contain predicates with verbs of giving and a postverbal noun which represents the indirect object and beneficiary.

Example (15) features another kind of postverbal noun which takes the form of a locative adjunct and represents the ‘place’ type of ‘disposal’ construction:

As Wu and Cao (2008) observe, the 跟 gēn object-marking construction also occurs with make verbs, the third type of ‘disposal’construction, mentioned above:

Hence, the construction subtypes correspond to those present in Medieval Chinese vernacular documents, according to Mei’s classification (1990).

Other types of postverbal nouns turn out to be possible as well, including ones representing the instrument, such as 布 pu²² ‘cloth’ in example (17), or ones coding a so-called ‘retained object’ in a part-whole construction (known as a băoliú bīnyŭ 保留 宾语) in example (18):

The ‘disposal’ construction in Waxiang also possesses the causative subtype:

Thus, the 跟 kai⁵⁵ construction can be seen to share a similar set of usages and subtypes to the ‘disposal’marker 把 bă in Mandarin and consequently to be a full-fledged construction in Waxiang in terms of its syntax and semantics.⁸ In the next section, we briefly outline similar functions for 跟 kε⁵⁵ in the Yuanling dialect of Waxiang to support our case.

3 The use of 跟 gen in the Yuanling Waxiang dialect

Yang Wei 杨蔚 (1999) was in fact the first person to describe the polyfunctionality of 跟 gēn in Waxiang, realised as 跟 kε⁵⁵ in the Yuanling dialect, the object of her study Its basic use is as a lexical verb meaning ‘to follow’ just as in the Guzhang dialect. For its extended uses, she presents fifteen examples from Yuanling Waxiang (1999: 190-191, 197), amongst which none, however, illustrate the comitative use (see §2.2.) or the locative source ‘from (a place)’. Apart from the use in Yuanling Waxiang to code a temporal source ‘from a point in time’, the rest of the functions are the same as those found in the Guzhang dialect for which we present the following seven examples.

3.1. The morpheme 跟 kε⁵⁵ can be used as a NP coordinative conjunction, just as in the Guzhang dialect (see §2.3.):

3.2. Just as in Guzhang Waxiang, 跟 kε⁵⁵ can also be used in the same way to introduce the direct object (see §2.8):

3.3. 跟 kε⁵⁵ is also used as a dative marker (see §2.5.):

3.4. 跟 kε⁵⁵ can be used as a benefactive (see §2.4.):

3.5. As in the Guzhang dialect described in §2 above, 跟 kε⁵⁵ can similarly be used with a following locative NP to introduce the direction of an action (see §2.6.):

3.6. Finally, 跟 kε⁵⁵ also has the meanings of ‘from’ or ‘since’ when used before a temporal expression indicating a reference point in past time (compare with §2.7.):

This concludes the comparison of the uses of 跟 kε⁵⁵ in Yuanling with 跟 kai⁵⁵ in Guzhang, two varieties of Waxiang. Analysis of the main issues begins in the next section wherein the relationship between the different functions of 跟 gēn is examined.

4 Lexical and grammaticalized functions of 跟 gēn in the structure [kai + np + vp]

Summarizing the discussions in the above section, the following eight different uses of 跟 gēn can be found in the Waxiang dialects spoken in Yuanling and Guzhang. Apart from the lexical verb usage, the remaining seven uses all occur in the structure that has the configuration (np) – gen – np – vp:⁹

1. a verb meaning ‘to follow’

2. a comitative meaning ‘along with (someone), with’

3. a locative meaning ‘along, towards (a place)’

4. an ablative meaning ‘from (a place), from or since (a reference point in past time)’

5. a NP coordinative conjunction meaning ‘and’

6. a dative marker meaning ‘to (someone)’ used with verbs of giving and verbs of communication

7. a benefactive marker meaning ‘for’

8. a marker which introduces the direct object

The first question to be discussed based on the above uses is: Are all the uses directly derived from the verb ‘to follow’? The answer is in the negative. Although some of them are directly derived from the verb, others are only indirectly linked to the meaning of ‘follow’. We classified the seven non-verbal usages into three main categories – adjunct prepositional phrases, conjunctions and oblique roles, each of which is now discussed in turn:

4.1. Adjunct prepositional phrases

Three prepositional uses, similar in function to oblique cases, are directly derived from the verb ‘to follow’. These are the three uses which form an adjunct phrase preceding the main verb:

1. the comitative ‘along with, with’

2. the locative/directional meaning ‘along a place, towards a place’

3. the temporal meaning ‘from, since’ and the source locative meaning ‘from’

According to regular patterns of semantic change involved in grammaticalization, the comitative preposition ‘along with, with’ has directly evolved out of the verb ‘to follow, to accompany’ in the Waxiang dialect. In the following examples, 跟 gēn in (27) is a lexical verb ‘to follow, to accompany’, as it is followed by the aspect marker tau⁴¹, while 跟 gēn in example (28) is treated as a comitative preposition ‘with’ in which traces of the original verbal meaning of ‘follow’ are nonetheless still apparent. This phenomenon has been explained in terms of ‘persistence’ by Hopper (1991).

The grammaticalization of the verb 跟 gēn ‘follow’ into a comitative preposition ‘along with, with’ is also found in Mandarin Chinese. According to Peyraube (1996: 191), the first real instances of 跟 gēn as a comitative preposition are found in the 18th century text 红楼梦 Hóng Lóu Mèng (Dream of the Red Mansions), despite the fact that some rare and scattered examples can be found earlier towards the end of 16^th century (see also Yu Jiang 1996, Feng Chuntian 2000: 301-323, and Wu Fuxiang 2003b).

However, in standard Mandarin, the usage of 跟 gēn is limited to ‘along with, with someone or something’, apart from its use to mark the addressee with verbs of communication,¹⁰ while in the Waxiang dialects, the verb 跟 gēn has developed further, adding on a pair of new meanings, these being (a) follow + Locative NP > ‘along, towards, to a place’ in the case of locative nouns (see example 8); and (b) follow + Temporal/ Locative NP > ‘from (a place)/ since (a reference point in past time)’, that is, an ablative sense (see examples 9 and 26 respectively). These semantic changes can be depicted in the following manner:

Figure 1.

Pathway of grammaticalization for 跟 gēn as a marker of adjunct PPs

It is interesting to observe that in Heine and Kuteva’s crosslinguistic study of grammaticalization, only the first pathway is exemplified for the verb ‘to follow’ and for just two languages, these being Ainu and Mandarin Chinese (2002: 139-140).¹¹

4.2. NP coordinative conjunction derived from the comitative preposition

The development from a comitative preposition into a NP coordinative conjunction, N-conj-N, is a well-known phenomenon in the history of Chinese. Liu and Peyraube (1994) showed that the Chinese coordinative conjunctions 及 jí and 与 yǔ (Archaic Chinese), 共 gōng, 同 tóng, 和 hé and 跟 gēn (Medieval and Modern Chinese) are not derived directly from verbs but from comitative prepositions, which have arisen themselves from verbs. These all bear witness to the pathway: verb > comitative preposition > coordinative conjunction.

This further stage in grammaticalization can be found in texts from the 19^th century for 跟 gēn (Liu and Peyraube 1994), but much earlier for the other comitative prepositions of Medieval and Modern Chinese: 共 gōng and 将 jiāng were grammaticalized into conjunctions by the end of the Six Dynasties period (222-589 AD), 和 hé was grammaticalized during the Tang (7^th-10^th century) while 同 tóng underwent this process during the Song, 10^th-13^th century (see also Yu 1996, Wu Fuxiang 2003b, Jiang & Cao 2005).

As mentioned earlier, in the Waxiang dialect there is another coordinative conjunction 邀 iau⁵⁵ which has this function as its main role. According to our field research, 邀 iau⁵⁵ is used when the relationship between N₁ and N₂ is equal, if not neutral (see examples 4 above and 29 below).

Examples (29) above and (30) below suggest that Waxiang 跟 kai⁵⁵ may be still in the process of developing from a comitative to a coordinative conjunction for noun phrases, since it is not as unrestricted in usage as 邀 iau⁵⁵. Possibly, a trace of its earlier meaning of ‘to follow’ remains, since kin terms may only be conjoined by kai⁵⁵ where the referent of N₁ belongs to the younger generation, such as a child, and N₂ refers to a relation from the ascendant generation, as we saw in example (2) above: ‘She went to buy fresh food with her mother.’ On the other hand, 邀 iau⁵⁵ is more common as a NP-conjunction with kin of the same generation. A similar situation pertains in standard Mandarin for 跟 gēn as opposed to 和 hé. Finally, note that these conjunctions are not used to conjoin verb phrases or clauses. – They are purely NP coordinative conjunctions.

As a comitative preposition meaning ‘with’, Waxiang 跟 kai⁵⁵ typically links two human NPs together: np_hum + preposition_com + np_hum. As a coordinative conjunction for NPs, it loses this semantic constraint regarding a [+human] noun and is able to conjoin any kind of NP, as example (30) demonstrates. At the same time, the three associated constituents, NP com NP, are reinterpreted syntactically as belonging to one and the same complex noun phrase, NP conj NP (for more details on the syntactic reinterpretation, see Stassen 2000 and Lai 2003a, 2003b on Northern Sixian Hakka 㧯 lau²). These features and stages are represented in Figure 2 below:

Figure 2.

Syntactic reanalysis of the verb ‘follow’ into a comitative preposition (Stage 1) and then into a NP coordinative conjunction (Stage2).

4.3 Benefactive, dative and object markers derived from comitatives

According to Heine & Kuteva’s lexicon of grammaticalization (2002: 139-140) which is based on material from approximately 500 languages including Mandarin Chinese, the verb ‘to follow’ can develop into a comitative as it does in Mandarin and in Waxiang. According to the same survey, the comitative can then further develop into (1) a marker of the agent in passives; (2) a NP-conjoining conjunction ‘and’ (3) a sentence-conjoining conjunction ‘and’; (4) a continuous aspect marker; (5) an existential marker; (6) a marker of instrument; (7) a marker of manner; (8) a passive marker in impersonal passives; (9) possessive ‘have’ and (10) a marker of temporal clauses (2002: 79–89). Significantly, benefactive, dative and direct object markers do not figure in their list as a target stage in grammaticalization for comitatives.

In fact, in Heine & Kuteva’s survey, benefactives are commonly derived from the verbs ‘to come to’ or ‘to give’ (2002: 73–74, 149–151), while datives are derived from allatives, benefactives and give verbs (2002: 37–38; 54, 153–154), and causatives, the closest semantically to what are known as the object-marking constructions in Sinitic languages, are derived from the verbs ‘to do’, ‘to make’, ‘to give’, ‘to take’ (2002: 117–118, 152, 286). Hence, what precisely are the mechanisms involved in the grammaticalization pathway of the comitative 跟 gēn into a direct object marker in the Waxiang language? In §6, we will lay out our claim that the object marker 跟 gēn is derived from the comitative via the benefactive and its closely related dative uses, suggesting that this contributes a new pattern to the stock of grammaticalization pathways, specifically, comitative > dative/ benefactive > object marker.

With regard to the pathway of benefactive > dative, we have not found any clear derivation of this kind in the entire history of the Chinese language, in spite of the cases listed by Heine & Kuteva (2002). Rather, evolution in the opposite direction from dative to benefactive is regularly attested. For example, the case of 以 yĭ: it is attested as a comitative preposition ‘with’ at the beginning of the Western Zhou (1066-771 BCE), then came to be used as a dative ‘to’ by the middle of this period but only as a benefactive ‘for’ at the very end of the Western Zhou (see Djamouri 2009). In fact, it is not always easy to distinguish the two (‘He bought me a pair of shoes yesterday’ = ‘He bought a pair of shoes for me yesterday’).

In the Waxiang dialects, there are four different orders for dative constructions which closely resembles the situation already in place for dative sentences in the Medieval period (see Peyraube 1986, 1988):¹²

Note that the construction in (34) is only attested in Mandarin Chinese and in Medieval Chinese with certains verbs such as 买 măi ‘to buy’, 写 xiĕ ‘to write’, etc. See Zhu Dexi (1979) and Peyraube (1988).

Although there are four types of dative structures, the most commonly used in Waxiang is: Dative preposition + IO + V + DO. This is unusual compared with the history of Mandarin, where the preverbal dative PP is the least used of the four structures just exemplified. Note that the examples above show that either the IO or the DO can be introduced by 跟 gēn and that either of these constituents can be located before the verb.¹³

The third Waxiang structure [跟 kai⁵⁵ preposition + IO + V + DO] introducing an oblique NP in the preverbal position, either in the benefactive or the dative role, is thus the most likely source of the object construction, as we will try to argue in §6 and §7. It represents the syntactic bridging context (Heine 2002) that prepares the ground for reanalysis of 跟 kai⁵⁵as a direct object marker (om) and, consequently, the dative/benefactive prepositional phrase [跟 kai⁵⁵+ NP] as being composed of [om+np_do].¹⁴

We hypothesize that the grammaticalization pathway for this development is as follows:

Figure 3.

Grammaticalization pathway from comitative preposition to direct object marker

Crucial to this grammaticalization pathway are the syntactic and semantic changes which underlie it, triggering the development from a preposition that marks oblique roles in an adjunct phrase, including the benefactive and the dative, in other words, non-arguments of the clause, to a fixed morphological marker introducing the argument of the direct object in a SOV clause.

5 Object markers and the comitative in Chinese dialects

As we lack historical documents for most of the Chinese dialects, we can only put forward the hypothesis given above concerning the grammaticalization pathway from a comitative preposition into an object marker, then back this up with evidence from dialects of Southern Min for which historical documents from 16^th-17^th centuries have been preserved (see §6). First, however, we would like to make the following three remarks:

1. The development of the comitative into an object marker, as found in Waxiang, is rare in the other languages of Hunan. For example, 跟 gēn as an object marker is not attested in any of the Xiang dialects, even though some of its other grammaticalized functions can be identified in the Southwestern Mandarin spoken in this western region of Hunan (see §5.1 below).

2. In contrast to this, 跟 gēn used as an object marker can be found in the Wu dialects spoken in Northeastern Jiangxi and also in the Gan dialects in contact with these (see Hu Songbo & Ge Xin 2003: 242, 245). It is also used in this function in certain Mandarin dialects spoken in Hubei and Jiangsu.

3. This use of 跟 gēn found in Waxiang shares similarities to that of comitative 共 gòng in the Min dialects, the topic of description in §6 and also 㧯 lau² in certain Hakka dialects.

The first two points are discussed in this section.

5.1. The non-usage of 跟 gēn as an object marker in the languages of Hunan

According to Y. Wu’s survey (1999) on object markers conducted in 100 localities in Hunan, the 100 tokens identified derived from twenty different lexical sources. Seventy-four evolved from the meaning ‘to hold’, ‘to give’, fifteen from the meaning ‘to take’, eight from a variety of other meanings including ‘to help’, while the meaning of a final three is unknown. See the details below (Note that the figures in parentheses after the subsection title indicate the total number of localities for each particular form).

5.1.1. Object markers derived from the meaning ‘to give’ (74)

Although the original meaning of 把 bă is ‘to hold’, it is commonly used as a verb meaning ‘to give’ in the Sinitic languages of Hunan; in fact, 62/74 of the localities make use of this verb as an object marker. Certain other object markers also share the source meaning of ‘to give’, as is shown in Table 2 below:¹⁵

Table 2.

Give verbs as the source for object markers

Marker	Meaning and representative dialect	Number
把 pa316	‘to hold; to give’ (e.g. Xiang- Changsha)	62
珙 kei3	‘to give’, (e.g. Xiang-Shuangpai)	4
分 pai1	‘to divide; to give’ (e.g. Jiangyong local dialect)	1
得 te5	‘to gain; to give’(e.g. Gan- Changning)	2
挨 a1	‘to be close to; to give’ (e.g. Ningyuan local dialect)	2
tou5	‘to give’ (e.g. Yizhang local dialect)	1
oŋ1	‘to give’ (e.g. Xintian local dialect)	2
Total		74

5.1.2. Object markers derived from the meaning ‘to take’ (15)

The following words can similarly be used as both free lexical verbs and object markers. Some only have the meaning ‘to take’; others have several meanings, one of which is ‘to take’. These words are nonetheless grouped together on the assumption that the principal source meaning is ‘take’.

Table 3.

Take verbs as the source for object markers

Marker	Meaning and representative dialect	Number
担 tan1	‘to carry (on one’s shoulder); to take’ (e.g. Xiang- Longhui)	5
拿 na2	‘to take’ (e.g. Yizhang local dialect)	3
拿倒 na2tə	‘to take’ (e.g. Xiang – Loudi)	2
捞 lau1	‘to dredge (up); to take’ (e.g. Xiang-Changsha)	1
提 tio1	‘to carry; to take’ (e.g. Zixing)	1
码 ma3	‘to put; to take’ (e.g. Gan-Chaling)	3
Total		15

5.1.3. A residual group of verbs as the source of object markers (8)

This group of object markers is derived from meanings other than ‘to give’ and ‘to take’ (8) and represents a farrago of markers with no particular common denominator of meaning. Nonetheless, it should be observed that certain of these can be used as causative verbs and are also common sources for agent markers in the passive.

Table 4.

Other verbs as the source for object markers

Marker	Meaning and representative dialect	Number
帮 pa1	‘to help’ (e.g. Xiang-Chenxi)	2
摸 mo1	‘to touch’ (e.g. Guiyang local dialect)	2
等 ten3	‘to wait’ (e.g. Huarong)	2
	‘to let’ (e.g. Huarong)	2
Total		8

5.1.4. Object markers derived from unknown meanings (3)

A final and small residual group of verbs is listed in Table 5 where the source of the object markers has not yet been identified.

Table 5.

Object markers whose source is not identifiable

Marker	Meaning and representative dialect	Number
	(e.g. Gan-Suining)	1
	(e.g. Daoxian local dialect)	1
to1	(e.g. Xiang-Luxi)	1
Total		3

In summary, Waxiang 跟 gēn presents the only case amongst all the Sinitic languages of Hunan that we are aware of, of a comitative preposition evolving into a direct object marker. These include those dialects belonging to the Xiang, Gan, Southwestern Mandarin and Hakka groups.¹⁷ Even more telling is the fact that there is no single piece of evidence in the history of the Chinese language of a comitative developing into an object marker and exponent of the ‘disposal’ 把 bǎ type of construction. In this respect, it is noteworthy that object marking constructions are those which have been studied in the minutest detail for all stages of Chinese (Peyraube 1989, Feng 2000: 555 ff., F.Wu 2003a/2006, Cao & Long 2005, Jiang 2008 etc.).

As was mentioned earlier, most people who speak Waxiang also speak Southwestern Mandarin as their second language. It has been observed that when Waxiang people speak Mandarin, they use both 把 bǎ and 跟 gēn as object markers (see Wu & Cao 2008). The following examples are taken from a narrative on the topic of ‘How to make the dish of suān roù (sour-flavoured pork)’. The informant was asked to tell us the recipe first in Southwestern Mandarin (see example 36) and then in the Waxiang dialect (example 37).

‘As for sour pork, when (you’ve) purchased the meat and come back home with it, brought the meat home, after you’ve brought the meat back, you clean it and cut it up.’ (XGX 2006)¹⁸

‘As for sour pork, for (the dish of) sour pork, (you) buy pork meat. After (you’ve) come back with it, (you) use water to wash (it). After you’ve rinsed it, you cut it into pieces.’ (XGX 2006)

In the Southwestern Mandarin example, (36), four object markers are used, two of which are 把 pa⁴¹ and two, 跟 kən⁵⁵. In the Waxiang example, (37), three object markers are used, all of which are 跟 kai⁵⁵. Can the object marking use of 跟 gēn be considered an independent development for Waxiang, introduced into the local dialect of Southwestern Mandarin from their own language, or was 跟 gēn once widespread in use in certain non-standard varieties of Mandarin, later being replaced by 把 bǎ while being retained in Waxiang? In order to address this issue, let us first consider some data from the Southwestern Mandarin dialects of Hunan province as well as from other Chinese dialects, including Mandarin, spoken in neighbouring provinces.

5.2. Southwestern Mandarin as spoken in Hunan

We would like to point out that in some Southwestern Mandarin dialects spoken in Hunan, 跟 gēn can be used as both a benefactive and a dative marker, even though it does not have the object marking use in question. For example, in the Jishou 吉首 and Changde 常德 dialects of Southwestern Mandarin, 跟 gēn shares a similar set of uses as found in the Waxiang dialect, except, notably, as an object marker (Li Qiqun, 2002: 301; Zheng Qingjun 1999: 245–246). Two examples of 跟 gēn used as a benefactive and/or a dative respectively are given below for each dialect.

1. Jishou dialect (SW Mandarin, Western Hunan):

   

   

2. Changde dialect (SW Mandarin, Hunan):

   

   

It should be noted that the variety of Mandarin spoken in Guzhang county of Western Hunan is not likely to be a possible source for this use of 跟 gēn in Waxiang, since monolingual Southwestern Mandarin speakers prefer 把 bǎ as the main object marker, even though recognizing and accepting the use of 跟 gēn. Hence, in answer to the questions posed above, it is significant for our hypothesis that this marker is imported by the present-day bilingual group of speakers of Waxiang into their use of SW Mandarin (for a more detailed discussion, see Wu & Cao 2008).

5.3. Non-standard Mandarin dialects outside of Hunan province

Huang (1996:526) provides data on three non-standard Mandarin dialects which use 跟 gēn as an object marker: two Jiang-Huai Mandarin dialects, Shuyang 沭阳 and Huaiyin 淮阴, both located in Jiangsu province, and one Southwestern Mandarin dialect, Suixian 随县, located in the neighbouring province of Hubei. We use the Huaiyin dialect to exemplify the use of 跟 gēn as an object marker in (42).

Presumably, further investigation will unearth yet other Mandarin dialects which use 跟 gēn as an object marker. At present, however, there is insufficient evidence to support any hypothesis regarding this function of 跟 gēn as having once been widespread in non-standard Mandarin dialects of central China. Diachronic evidence also provides an answer to the contrary, as pointed out earlier in §4.1.

5.4. 跟 gēn used as an object marker in the Wu and Gan dialects of Jiangxi province

According to Hu Songbo & Ge Xin (2003: 243), this use of 跟 gēn as an object marker occurs in certain Wu and Gan dialects of northeastern Jiangxi province, although no detailed descriptions are provided. We present one example, taken from the Shangrao Wu dialect:

Hence, the use of 跟 gēn as an object marker in Sinitic languages appears to be scattered across a swathe of central China and affects a handful of Southwestern and Jiang-Huai Mandarin dialects as well as certain Wu and Gan dialects, located in the provinces of Hunan, Hubei, Jiangxi and Jiangsu. This area corresponds to a transitional zone for Chinese dialects between the large Mandarin group to the north, northwest and the southwest and the Southern Sinitic languages to the south and southeast.

The Hui dialects excepted – which generally make use of give and help verbs in this function (Hirata 1998, Chappell 2007) – it is interesting to observe that the Central dialects concerned here pattern in a similar way to many Hakka and Min dialects in Fujian and Guangdong provinces, not to mention the Taihu Wu dialects in Zhejiang province. And this is specifically with regard to having a comitative as the source of their object marker.

The general conclusion on this issue is therefore that the use of 跟 gēn as an object marker appears to be quite rare, as far as our present state of knowledge is concerned. In written Mandarin, it is attested as a comitative preposition only from the 18^th century or at the earliest from the 16^th century, as suggested by Yu Jiang (1996) and Wu Fuxiang (2003b), suggesting that its use as an object marker is likely to be a quite recent development in the non-standard Mandarin dialects where this is found, and possibly only from the 19^th century (see also §6 on similar developments for 共 gòng in Southern Min). In contrast to this, it is well-known that the 把 bǎ construction or the 将 jiāng construction became fully grammaticalized and was the predominant form during the the Tang dynasty (7^th – 9^th century) (Zhu Minche (1957), Wang Li (1958), Peyraube (1985, 1989). Competition with the prestige form of the court language, coded by 把 bǎ, would certainly have limited the impact and the diffusion of the 跟 gēn innovation across the regional varieties, as a function of the degree of contact with officialdom.

Surrounded by a largely non-Sinitic population, Waxiang speakers remained isolated and predominantly monolingual up until the latter half of 20^th century, as observed in §1. This may well have created the right conditions for the independent development of 跟 gēn into an object marker and its retention, despite the widespread diffusion of 把 bǎ throughout the core Mandarin-speaking areas of China.²¹ In this respect, consider Chen & Li’s survey of 96 Mandarin dialects (1996: vol 5) which includes all the main subgroups: they list only 4 as not having 把 ba as the primary object marker.

5.5. Comitative markers from other sources used as object markers

Apart from 跟 gēn, other comitatives can be found to function as object markers in the Southern Sinitic languages of Hakka, Wu and Min, which although they belong to the same semantic field, nonetheless have distinct, non-cognate sources. In Hakka, two common comitative morphemes are 同 tóng < ‘to be together’ and 㧯 lǎo ‘to mix’ (see Lin 1990, Lai 2003a, 2003b, and Chappell 2006) while in the Taihu group of Wu dialects in Zhejiang province, comitatives, including 則 tse?⁴⁵, have also been attested as object markers (see Xu & Tao 1998). In addition to this, a pan-Min feature appears to be the use of comitative 共 kā (or gòng in Mandarin) which shares many similarities in usage with Waxiang 跟 gēn.

Lai (2003a, 2003b) treats Hakka 㧯 lǎo in detail while Chappell (2007, to appear) discusses many of the markers mentioned above, showing that synchronically they are still used as comitatives. Hence, in the next section, we confine our discussion to just one of these sources, namely 共 gòng in Southern Min, in order to give some details on its diachronic development that may shed light on the grammaticalization processes involved in this unusual pathway and lend support to our hypothesis. We will also briefly discuss the usage of the same 共 gòng in the Fuzhou dialect (Northeastern Coastal Min) in the next section.

6 Diachronic syntax and semantics of the comitative in Southern Min: The grammaticalization pathways of 共 gòng

In modern Min dialects,共 gòng has many different usages, including the comitative, benefactive, dative and object marking ones, thus showing striking similarities with 跟 gēn in the Waxiang dialects, described in §2, §3, and §4, albeit the two have distinct sources. The lexeme 共 gòng is attested in Archaic Chinese where it meant ‘to be the same as’ and later ‘to share with’ and ‘to accompany’. It was probably only during the Six Dynasties period (222-589AD) that 共 gòng was grammaticalized into a comitative preposition (see §7 and Liu & Peyraube 1994, Yu 1996, Wu Fuxiang 2003b).

Contemporary Min dialects similarly continue to use cognate forms of 共 gòng as their main comitative markers, expressing the comitative ‘with’ and connective ‘and’ meanings. Relevant for the present analysis, they also use it as their object marker (for descriptions, see Chen 1998; Chappell, 2000; Lien, 2002 and Tsao 2003 among others).²² It could in fact be seen as a typological feature of the Min dialects, distinguishing them from other Sinitic languages, particularly in the case of Eastern or Coastal Min (see Norman 1987 for the classification of Min).

In this section, we examine data from historical materials on Southern Min from the early 17^th century, which clearly show the polysemy of the comitative marker 共 gòng.²³ Apart from Mandarin and written genres of Chinese such as Classical Chinese and old orMedieval vernacular documents, Southern Min presents a rare opportunity for diachronic research, since historical data is available that dates back to four centuries ago. Three main texts have been examined for this purpose. The first two represent the same dialect of Southern Min that was spoken by a sizeable community of southern Fujian traders who had settled in Manila by the late 16^th century. These documents use both Chinese characters and the romanization created by Spanish missionaries working in the Philippines:

1. Doctrina Christiana en letra y lengua china, (1607), Vatican Library. This is the translation of the Spanish version of the Doctrina Christiana into a Southern Min dialect²⁴

2. Arte de la lengua Chiõ Chiu [Grammar of the Chiõ Chiu language] (1620), University of Barcelona Library. Handwritten title: Gramatica China.²⁵

In addition to these missionary materials, we also make some reference to a third text from China:

• (iii)荔鏡記 Lì Jìng Jì [Romance of the litchis and the mirror], Jiajing edition (1522-1566, but written probably under the 14^th century Yuan), corrected and annotated by Wu Shouli (2001). This is the earliest version of a Ming dynasty play written in a mixture of two Southern Min dialects, Quanzhou and Chaozhou dialects.²⁶. We will also make reference to the second edition of Lì Jìng Jì, called Lìzhī Jì [Romance of the litchis], Wanli edition dated 1581.

En passant, fleeting mentions are made, where pertinent, to the manuscripts Dictionarium Sino-Hispanicum, believed to have been compiled in 1604 by Pedro Chirino in Cebu and the Bocabulario de la lengua sangleya, compiled in Manila ca. 1617 by Dominican priests.

The form 共 gòng corresponds to the function word câng in our two main historical documents. In the section on case marking in the 17^th century Arte, câng is described as having the same ablative meaning as cab 甲 and tāng 同 (1620: 3a, 3b) when preposed before a noun. Nonetheless, all three are glossed by Spanish con ‘with’, as in the example 共 人 câng lâng – con el hombre ‘with the man’. A slightly more extensive description found further on in the grammar is reproduced below (1620: 11a, 11b):²⁷

Câng 共 (Arte 1620: 11a, 11b)

Cang: esta particula. cang. tiene.particu-lar dificultad por.no(s).penetrar su significación, unas vezes sinifica con.i.sirve para ablativo co-me que.da. dicho, otras vezes es vervo y sinifica ayudar.

“Cang: This particle cang creates a special difficulty for us to plumb the depths of its meaning. Sometimes, it means ‘with’ and is used for the ablative, as earlier mentioned. Other times, it is a verb meaning ‘to help’.”

(…) Pero es de advirtir que.son.equi-bocas estas oraçiones puede açer sentido que me ayude en conprar pescado de otros para mi o que el pescado es mio que le conpre.

“However, it should be pointed out that these utterances are ambiguous. The meaning could be that someone helps me to buy fish from others for me, or else that the fish is mine and that someone buys it from me.”

De suerte que el cang junto con vervos de conprar o bender o prestar o ablar varia notablemente la sinificacion como esto chio. “with the result that cang along with verbs of buying, selling, lending or speaking noticeably changes its meaning, as with this chio ‘lend’.”

The explanation continues by pointing out the ambiguity of the verb 借 chio ‘lend’ which, when used with 共 câng, may mean either ‘to borrow’ or ‘to lend’.

This largely resembles the description for the cognate preposition found in the Carstairs Douglas dictionary compiled 250 years later, with respect to the closely related Southern Min dialect of Amoy (Xiamen), as recorded in the mid-19^th century:²⁸

kā (…) a preposition, used with persons only, meaning to, from, for: (…) kā-i-khui, to open it for him, as a door. (Douglas 1873: 188)

Douglas too, and Barclay in his supplement to Douglas‘ dictionary, both give examples of transactional verbs where the direction is ambiguous between the benefactive ‘for’ and ablative ‘from’ readings:

kā-i-thoeh, take or receive from him (Douglas 1873: 188) góa kā lí bóe, I will buy from you, or buy for you (Barclay 1923: 91)

The main manuscript examined, the Doctrina Christiana, contains a total of 33 examples of the morpheme câng, that is, determined as a function of the Spanish romanization.²⁹ As can be seen from Table 6 below, the primary uses of câng are those of a coordinative conjunction ‘and’ with 13 tokens, while the benefactive ‘for’ is the most prominent, amongst the oblique roles, with 12 tokens, followed by the ablative ‘from’ with five (5) and the dative ‘to’ with just two (2). There is only one example of a comitative ‘with’. No clear, unambiguous cases of an object marking function of câng are identifiable. However, it is extremely important to note that at least ten examples in the benefactive category are ambiguous, having the potential to be interpreted as patients as well, a point we elaborate upon below.

Table 6.

Distribution of meanings for cang in the Doctrina Christiana (ca. 1607)

Meaning	Tokens
1. Comitative ‘with’	1
2. Conjunction ‘and’	13
3. Dative ‘to’ (Goal)	2
4. Ablative ‘from’ (Source)	5
5. Benefactive ‘for’ (Beneficiary)	12
6. Object marker (unambiguous cases) (Patient)	0
Total	33

Note that all these uses are also present in contemporary Taiwanese Southern Min for the marker 共 kā, according to Cheng & Tsao (1995) and Tsao (1991, 2003): 共 kā can introduce Goal ‘to’ and Source ‘from’, while it also has the benefactive ‘for’ and comitative ‘with’ uses. In addition, it can introduce the direct object with monotransitive verbs, while it can also act as a prefix on transitive verbs.

Furthermore, all these different uses of 共 kā invariably have the structure:

(np_agent) – [共 kā – n] – vp

Let us now examine these uses one by one.

6.1. Comitative marker

In the Doctrina Christiana, there is just one clear example of the comitative ‘with’ use that is used with the verb 辯論 pien lun ‘to debate, argue with’:³⁰

The low frequency in the main text of this usage, compared with the more grammaticalized usage as a NP conjunction (§6.2.) suggests, albeit weakly, that by the time of Early Modern Southern Min (late 16^th century – early 17^th century), the less grammaticalized uses, such as this one, were already on the wane. This is corroborated in terms of its endpoint by Lien (2002: 195) who observes that the comitative function of 共 gòng, (i.e. câng), also evident in the Lì Jìng Jì, has undergone lexical replacement in all four major dialect groups of Southern Min today: Jieyang, Longxi, Xiamen and Jinjiang.

In the Lì Jìng Jì (Jiajing edition of 1566, but probably written in 14^th century), the comitative function of 共 gòng is, on the contrary, very frequent: there are 144 comitatives ‘with’ ; 61 datives ‘to’ (most used with verbs of saying) ; 41 coordinative conjunctions ‘and’ ; 40 benefactives ‘for’ ; 8 ablatives ‘from’ in more than 300 occurrences of 共 gòng (the others being, above all, the adverb ‘together’)

The same situation is found in the second known version of the same opera (Lìzhī Jì, Wanli edition of 1581) : out of almost 250 共 gòng, there are 104 occurrences of comitative ‘with’ ; 43 occurrences of benefactive ‘for’ ; 42 occurrences of the conjunction ‘and’ ; 29 occurrences of the dative (goal) ‘to’ which are all used with verbs of saying; and 3 occurrences of the ablative ‘from’.³² In neither of these two editions is 共 gòng used as a verb meaning ‘to gather, to share’ or ‘to help’.

6.2. NP coordinative conjunction

This group forms by far the largest category of examples in the Doctrina with 13 occurrences of câng in this function in the text. Câng in each of the 13 examples is used to connect equal status nouns and these mainly denote, or are associated with, human referents. Example (45) is representative of this:

There are also four examples of the conjoined NP ‘body and soul’ occurring in different sentences in the Doctrina Christiana:

6.3. Dative marker ‘to’

In the Doctrina Christiana, there are two examples where câng introduces the addressee with verbs of communication and where the Goal sense of ‘to’ is coded. There are, however, no prototypical examples with verbs of giving.

The following three points provide further evidence of this tendency for dative câng to be used with verbs of saying: First, 共 câng also occurs twice in the Arte with the speech act verbs 呾 taⁿ ‘to talk’ and 說 suè ‘to discuss’:

Second, more than half the examples of câng in the Dictionarium Sino-Hispanicum are combined with verbs of saying (7/12). Third, in the Lì Jìng Jì, verbs of saying occur with the highest frequency in this category with 共 câng. In the Jiajing edition, 58/61 clauses with dative 共 câng are used with this verb class, including the same two main verbs, 呾 taⁿ ‘to talk’ and 說 suè ‘to discuss’ as above, and ‘to answer’, ‘to ask’ and ‘to discuss’ (see also Lien 2002 for similar remarks). The term ‘dative’ is thus used in its extended sense, since câng in 16^th century Southern Min is clearly not used as the preposition for dative constructions of transferral where the main verb is a verb of giving. This is the province of 乞 khit < ‘give’ (see Chappell 2000, Lien 2005).

6.4. Ablative marker with verbs of taking away

In this use, 共 câng marks the source in a similar manner to an ablative case marker with verbs of deprivation or taking away, such as ‘buy’, ‘borrow from’ and ‘cheat (of money)’. This corresponds to the sense ‘from’ mentioned in the Carstairs Douglas dictionary (1873: 188):

In the Doctrina, there were five ablative uses of 共 câng, all with the verb kiu ỹin (xin) cheng 求人情 ‘beseech mercy (from God)’, which belongs to the semantic field of ask verbs, an extended sense of the ablative (‘to ask something from someone’).

In the Jiajing edition of Lì Jìng Jì and in the Wanli edition of Lìzhī Jì, we also have a few examples of this ablative use ‘from’ with verbs such as ‘buy’, ‘separate’, ‘receive’, ‘ask’.

This, like the dative use of câng, proves to be quite restricted in the range of verb classes it combines with across the 16^th-17^th century texts consulted.

6.5. Benefactive marker

The benefactive use forms the largest category among the three oblique roles coded by 共 câng and is evident in 12 occurrences in the Doctrina Christiana, including the two examples below.

The benefactive use of 共 câng appears to be less restricted in its distribution than either of the ablative and dative uses, since it is found with a large range of predicates including different verb classes (for a list, see Lien 2002).

6.6. Object marker with monotransitive verbs

Several recent treatments of 共 gòng in Min dialects show that it has a similar function to the 把 bă construction in Mandarin where it marks a preverbal and typically referential direct object and co-occurs with a transitive action verb; for example, see Cheng & Tsao (1995), Chappell (2000) and Tsao (1991, 2003) on Southern Min and Chen (1998, 2006) on the Fuzhou dialect of Northeastern Coastal Min. A relevant example follows, taken from a modern Taiwanese Southern Min story:

The use of 共 câng to unequivocally mark a patient was not found in either of the early 17^th century missionary texts, nor in our two versions of the Lì Jìng Jì.³³ Neither was it present in the related Fuzhou dialect before the first half of the 20^th century in the case of its cognate marker, according to Chen (2006) (see below for a rapid sketch on the uses of 共 gòng in the Fuzhou dialect).

In the second half of the Doctrina, which contains a larger number of Classical Chinese elements, we do find, however, two examples of 將 jiāng used in this object-marking function where it combines with literary expressions such as the following patient NP 他手足 t’a siu chioc ‘his hands and feet’. This marker is romanized as chiang in the Doctrina:

Note that many Southern Min dialects use a cognate of this object marker 將 chiong¹ or chiang¹, which belongs to the literary stratum, alongside the use of a colloquial level marker, typically a cognate of 共 gòng (see Lien 2002, Chappell 2006, 2007 and Table 2 in Chappell 2000 for Min dialects and their use of 共 gòng and 將 jiāng). By way of contrast, in the Ming dynasty play, Lì Jìng Jì, the object marking construction is typically marked by either 力 liah⁸ (a demotic character for 搦)or 將 chiong¹, according to Lien (2002). Interestingly, some examples also make use of another comitative marker,甲 kah⁴:³⁴

The lack of an object-marking function notwithstanding, this use of 共 gòng [kang⁷] certainly could be a contextual inference in some of the examples in the Doctrina Christiana, such as in (53) above, even if it had not developed into an invariant component of meaning in this subtype in early Southern Min. In (53), the priest is not only performing one of the seven important sacraments of the Catholic church for the benefit of the believer, but is also doing the action to him (or her), that is, lang ‘person’ is simultaneously the beneficiary and the patient who is undergoing the rite of baptism.

Liu Danqing (2003) has similarly observed that with certain verbs, particularly the verb 帮 bāng ‘help’, the patient and the beneficiary have a tendency to be confused. This situation corresponds to the ‘bridging context’ defined by Heine (2002), whereby a new inference concerning grammatical meaning becomes contextually available, enabling a simultaneous interpretation as a patient in the present case, while not cancelling out the beneficiary sense.

In another example of an apparently benefactive use, the priest carries out the sacrament of Extreme Unction and anoints the believer with Holy Oil. The referent of the noun marked by câng, y ‘3sg’, an anaphoric pronoun for the preceding lang ‘person’, is thus both the beneficiary of the blessing and the patient who is physically anointed with the Holy Oil.

Ten of the 12 benefactive examples show such ambiguity with a second possible patient reading for the noun following 共 câng. For example, several predicates whose actions have 巴禮 Pare ‘priest’ as the agent, include ‘perform marriage’, ‘baptize’, ‘absolve someone (of their sins)’ as well as ‘anoint with Holy Oil’, the latter exemplified directly above. In all such cases, the priest is both performing the action for the person as well as doing it to him/her, thereby causing a direct effect.

In conclusion, the usage of 共 gòng as a comitative and a conjunction in modern Southern Min dialects such as Taiwanese shows, on the one hand, a preservation of a feature of Medieval Chinese grammar in Southern Min. Its range of functions, on the other hand, has increased significantly to include the object-marking one as a principal use.

In the contemporary dialect of Fuzhou, the use of 共 gòng as an object marker is also found. According to Chen Zeping (2006), 共 gòng is used in the Fuzhou dialect as (i) a verb (very rare), with the meaning of ‘have relations’, ‘take care’, (ii) a coordinative conjunction ‘and’, (iii) a dative preposition ‘to’, (iv) a comitative preposition ‘with’, (v) an ablative preposition ‘from, since’, (vi) a benefactive preposition ‘for’, (vii) an object marker. In short, all the same uses of 跟 gēn in Waxiang or 共 gòng in Southern Min turn up again.

Chen also notes that in grammars and dictionaries of the Fuzhou dialect redacted by missionaries in 19^th and early 20^th centuries, the uses of 共 gòng as a comitative preposition, as a benefactive preposition, and as a coordinative conjunction are found, but it never turns up as the object marker in a ‘disposal’ construction. This last use as an object marker thus appears to be very recent, even more so than in Southern Min.

In the final balance, he proposes that the object-marking use of 共 gòng does not directly evolve from the verb 共 gòng (which in any case had become quite rare), or from the comitative preposition, but from the benefactive preposition. Chen does not however provide any arguments for this. While his hypothesis is clearly the same as ours, we set out to justify our position in the following sections by furnishing many more examples than Chen, and by analysing them.

7. An explanation of diachronic change for 共 gòng and 跟 gēn

The lexical source for 共 gòng can be traced back to a verb ‘to gather’, ‘to share (with)’ as used in pre-Qin Classical Chinese (5^th – 2^nd centuries BC). By the end of Late Archaic (2^nd century BC), this verb had developed a grammaticalized function as an adverb ‘together’, the semantic input to its later uses of the preposition ‘with’ and the conjunction ‘and’.

According to Liu and Peyraube (1994: 187-188), the use of 共 gòng as a comitative preposition ‘with’ is attested from the Early Medieval period (2^nd – 6^th centuries) onwards, while its use as a conjunction ‘and’ became common in the second half of the Late Medieval period, specifically from the Song dynasty onwards (10^th-13^th centuries); cf. also Liu Jian (1989).³⁵

As a lexical item, 共 gòng continued to be used as both a preposition and conjunction until the end of the 16^th century in Early Mandarin, after which it was supplanted by 和 hé in Northern Chinese (see Liu and Peyraube 1994 for details). The functions of 共 gòng (câng) in the three 16^th and 17^th century Southern Min texts thus clearly maintain most of those found for Medieval Chinese 共 gòng used as a preposition and a conjunction.

Below is a time-line summarizing the description given in Liu and Peyraube (1994) for the development of the different grammatical functions of 共 gòng which we next compare with the situation for Waxiang 跟 gēn.

Figure 4.

Timeline for 共 gong

7.1. Grammaticalization pathways affecting comitatives

By the period of Early Modern Southern Min, it is clear that both the comitative preposition and coordinative conjunction uses are still available for 共 gòng, according to our historical materials discussed in §6 above. That is, Early Southern Min either inherited the uses of 共 gòng from some ancestral form of the Chinese language, the same uses available in Medieval Chinese or else, at a period prior to the time of our texts, grammaticalization into a conjunction had occurred independently.³⁶

An important contrast between Late Medieval Chinese and Early Mandarin for the use of 共 gòng is that, in Mandarin, the steps in grammaticalization to benefactive, dative, ablative (source) and object-marking uses are not found. How did these pathways for 共 gòng in Min eventuate?

We would like to propose that the comitative use is the launching pad for not just the NP coordinative conjunction ‘and’, as is well-attested in the history of Chinese and also for many languages in the world (Stassen 2000, Stolz 2001, Heine & Kuteva 2002) but also for an oblique marking function, that is, that there are at least two pathways of grammaticalization in the Min dialects that evolve from the comitative use. The second, which is less well-known in the typological literature, is this extension from the accompaniment sense of ‘with’ to marking non-argument roles such as the oblique functions of the dative and the benefactive, the addressee or the ablative. This can be seen as cases of context-induced grammaticalization (Heine 2002). Verb classes determine which particular role the NP following the marker will assume (speech act, transactional, deprivation verbs etc). This stage could have created the right environment for the development of an accusative or object-marking function to subsequently take place, the final stage which is discussed in more detail below. A brief overview of the scant literature on this grammaticalization pathway is next presented before setting out justifications for our model.

For the first stage in the grammaticalization chain regarding the change from comitative > oblique argument, this model finds a certain justification in the evolution of the preposition 以 yĭ of Classical Chinese (Archaic period). As Djamouri (2009) has shown, this preposition, arisen from the grammaticalization of a verb meaning ‘to lead’, is first of all attested as a comitative preposition (in the first period of the Western Zhou, 1066-771 BC), then later, in the final period of the Western Zhou, as a benefactive/dative, locative or directional preposition.

In addition to this, an analysis by Liu Danqing (2003) highlights two coordinative conjunctions 搭 dá and 帮 bāng in Northern Wu dialects (Shanghai, Wuxi, Suzhou) which also serve as comitative and benefactive prepositions. For the first (搭 dá), he proposes two pathways of grammaticalization : (i) verb with the meaning of ‘join, ‘connect’, ‘be together’ > coordinative conjunction; (ii) verb > comitative preposition ‘with’ > benefactive preposition ‘for’. For the second (帮 bāng), however, he proposes one single pathway of grammaticalization, i.e. verb with the meaning of ‘help’ > benefactive preposition > comitative preposition > coordinative conjunction. The benefactive preposition 帮 bāng is in effect prior in time to the comitative preposition and the conjunction in its development. This proposal with its two opposing sets of derivations – benefactive > comitative and comitative > benefactive – for the same group of dialects is, as a consequence, very intriguing but not fully convincing, as a benefactive > comitative derivation would violate the unidrectionality principle of grammaticalization (the comitative being less grammaticalized than the benefactive). In fact, solid diachronic data is lacking that would allow us to accept what is quite an an iconoclastic hypothesis.

For the final stage in the proposed grammaticalization chain from benefactive/dative > accusative, Tsao (1991: 382–385) claims that the Source or ablative role is the most likely basis for the reanalysis of Southern Min 共 gòng as an object marker. In contradistinction to Tsao, for the case of 㧯 lau in Northern Sixian Hakka of Taiwan, Lai (2003b: 552-555) argues that the patient function is a semantic extension from either the source or ablative function with verbs of deprivation or from the benefactive, specifically its malefactive counterpart.

We would like to maintain our alternative proposal which purely takes the benefactive/dative role as the point of departure for the transition to an accusative role, following Chappell (2006, 2007). Note that Endo (2004) has also suggested this pathway in a general way for Wu and Min dialects, as too Chen (2006) for the Fuzhou dialect of Northeastern Min (as observed already in §6.6 above). However, neither analysis elaborates on their position. The new hypothesis is as follows:

Figure 5.

Grammaticalization pathway for comitatives which evolve into object markers

This change occurs specifically via syntactic reanalysis of the oblique benefactive/dative role as a patient, entailing a dramatic reinterpretation of a non-core oblique argument as a core argument. The pathway is not unique to Sinitic languages with respect to the stages benefactive/dative > accusative. Heine & Kuteva (2002: 103, 37) also identify cases of the related dative case developing into an accusative marker, namely, the example of contemporary Spanish, where animate nouns may be marked as direct objects with the normally dative case preposition, a.³⁷ They also describe a diachronic change that occurred between Old English and Modern English for him_dat > him_dat/acc and hire_dat > her_dat/acc, involving the reanalysis of former dative case pronouns as both accusative and dative.The same kind of scenario is claimed to exist for Persian râ (Heine, Claudi and Hünnemeyer 1991: 165-168).

Since, as we have shown from our data, the dative use of these former comitative markers in Min dialects is rather limited, we merge it with the function of the benefactive ‘for’ in the following model which we propose to account for these developments in Min.

Figure 6.

Grammaticalization pathways for 共 kā ~ kāng in the Min dialects

We could also have formulated a completely different hypothesis consisting in deriving the accusative marker 共 gòng directly from a verb, namely a verb meaning ‘help’ which is attested in the Arte (11a, 11b), (and not from the verb ‘be together, share’): 汝共我買魚 lù câng gùa bèi hū ‘You help me to buy fish’. Indeed, as observed in the introduction, verbs of helping are one of the three main sources of object markers in Sinitic languages. This hypothesis does not however seem reasonable to us in the present case, given that the appearance of this object marker in Min is very late, and during a period when the verb 共 gòng meaning ‘help’ was doubtlessly no longer attested.

In fact, we can even doubt that this meaning was ever attested for Southern Min. In the different versions of the Lì Jìng Jì, there are, in fact, no examples of 共 gòng = to help. The Spanish translation of ayudar found in the Arte is thus somewhat controversial.

Could it be then a case of exaptation? Exaptation refers to the situation where a form has lost its function, or has become completely marginal within the system, and so is available for reuse in some other new function. Our view is that the grammaticalization of 共 gòng certainly does not meet this description, given that exaptation involves reuse of an old form, A, for something else completely new, (B), without any direct or indirect connection between A and B (Peyraube 2007).

There is also a semantic change involved in grammaticalization from an oblique marker to an object marker: A further semantic extension to marking the direct object is suggested by particular examples whereby an action performed for someone (beneficiary) can be reinterpreted as an action performed on someone (patient), cf. the example of a baptism. As pointed out above, the priest is not only doing something for the said person (baptizing them so that they can become Christian) but is also doing something to them (sprinkling water on them). This provides a bridging context for the pathway to the final object marking use in which the beneficiary sense no longer finds any interpretation.

This turning point in grammaticalization or ‘switch context’ occurs when the interpretation of an accusative or object marking role excludes all others (Heine 2002). Significantly for our analysis, this semantic change is context-induced, as we have set out to explain above. The original comitative meaning of these different markers meaning ‘with’ thus becomes incompatible with the new grammatical meaning, the only one now available for the construction. This is accompanied by conventionalization of the marker’s use such that the clause is syntactically incomplete if the [Object Marker – NP] constituent is omitted. In other words, the erstwhile comitative preposition undergoes obligatorification as a new morphological marker of the direct object, thus entailing syntactic renanalysis of its constituency. A remarkable consequence of this is that it is now able to mark a core argument of the verb, licenced precisely by this context-induced semantic change: S – ([Prep -N]_pp) – VP > S – [Object marker – N_do]_np – VP.

The transition is not an abrupt one and may proceed by means of diffusion throughout different lexical categories of nouns acting as the direct object, beginning with those having animate referents, just as in the case of the Spanish dative > accusative reanalysis. In fact, just such a description has been proffered by Lien (2002) who claims that the object nouns marked by 共 gòng in Southern Min do not represent prototypical patients. This may certainly have been the case in the early use of the object marking construction, for which we have, regrettably, no data. However, according to our database of contemporary Taiwanese Southern Min materials, this semantic constraint no longer holds and inanimate NPs may serve equally well as the object of 共 gòng.³⁸

Given that we lack any historical documentation for Waxiang, we propose, by means of comparison with the related language of Southern Min, that a similar series of grammaticalizations could have taken place in Waxiang, beginning with the comitative stage.³⁹

Figure 7.

Grammaticalization pathways for 跟 gēn in the Waxiang dialects

Conclusion

In summary, we have set out to demonstrate that the comitative 跟 gēn in the Waxiang dialect shares a set of very similar usages to 共 gòng in the Min dialects, despite their distinct lexical sources. Specifically, they act not only as comitative prepositions as in Mandarin, but also, quite unusually, as direct object markers, among a range of other functions. To this end, we have proposed a series of grammaticalization chains to account for the stages in development from comitative to accusative, that is, the direct object marking function.

The attempt to motivate this grammaticalization chain has made recourse to the principle of context-induced grammaticalization, entailing semantic change that has resulted in a major syntactic reanalysis of a non-core oblique NP into a core argument of the VP. We have furthermore conjectured that this cannot proceed by a single and direct step in grammaticalization but must involve the specific intermediate stage of an oblique role interpretation as either benefactive or dative. It was concomitantly argued that this grammaticalization chain has to be separate from the pathway leading to the emergence of a coordinative conjunction.

A similar set of developments can be hypothesized for Southern Min for which a variety of historical documents dating from 16^th-17^th centuries is serendipitously extant, and which illustrates a comparable set of polysemous uses, except for the object marking one. This appears to support our contention that an object marking stage is the most recent one in the chain of case-like roles.

Finally, we suggest that the use of 跟 gēn as an object marker in other Mandarin dialect subgroups, including Jiang-Huai and Southwestern, merits further research. Equally important would be to carry out a crosslinguistic survey to investigate if there are further cases of this unusual pathway of grammaticalization from a comitative to an object or accusative marker via the benefactive/dative stage in other languages of the world.

Grammatical abbreviations used in the glossing

cl

classifier

com

comitative

comp

verb complement

conj

conjunction

dat

dative

dir

directional complement

dat

dative

do

direct object

gen

genitive

hum

human

io

indirect object

mod

modal particle

neg

negative

nom

nominalizer

NP

noun phrase

om

object marker

pfv

perfective

pl

plural

prt

aspectual particle

rdp

reduplication

sg

singular

sp

structural particle

v

verb.