In this brief response to Jeff Siegel's paper (Siegel in press, this issue), I wish to highlight two issues; one is conceptual, and the other is empirical. I will address the conceptual issue first.

It is well-known that grammatical features transfer from the linguistic substratum to the contact language, be it a pidgin/creole or a new variety of English. How the transfer takes place is not a straightforward matter in contact linguistic theorizing. The fundamental question is the target of transfer: does transfer target grammatical systems/constructions or surface morphosyntactic strings? According to the congruence constraint (Siegel Reference Siegel1999, Reference Siegel2008), transfer targets actual word order – the ‘leaves’ of the syntactic tree. For a given substrate grammatical function, mismatch in surface position between the substrate morpheme and its putative counterpart in the lexifier language severely curtails transferability of the said function. In a series of papers on Chinese-derived grammatical features in Singapore English, I argue for a systemic approach to substratum transfer (Bao Reference Bao2005, Reference Bao2009, Reference Bao2010; Bao & Lye Reference Bao and Lye2005). The basic idea is this. The entire grammatical system, such as the aspectual system, is targeted for transfer. The constructions that comprise the transferred system are exponenced, or ‘fleshed out’, with the appropriate morphosyntactic materials from the lexifier language. For a given construction, the surface position of the exponent in the lexifier language need not match that of the exponent in the substrate language. It is the lexifier language that ultimately determines the position of the exponent in the contact language. Using the experiential aspect ever as an example, the two models of substratum transfer can be schematized as follows (H=Hokkien; M=Mandarin; SE=Singapore English):

1. (1)

   1. (a) Congruence-based transfer (Siegel Reference Siegel1999, Reference Siegel2008 )

      

   2. (b) Construction-based transfer (Bao Reference Bao2005, Reference Bao2009)

      

      M: V guo H: V-ke, bat V, bat V-ke SE: ever V

In (1), the target of transfer is boxed. For ease of exposition I use the term Hokkien more narrowly, to refer to the Southern Min dialect spoken in Malaysia and Singapore. The Hokkien and Mandarin forms will be discussed shortly.

The two models offer different perspectives on substrate influence, with serious empirical consequences. Siegel mentions two, the experiential and the imperfective. We will consider the experiential first. Siegel's model, as we can see in (1a) above, requires the preverbal bat as the source of the experiential function of ever in Singapore English. While it is true that Southern Min generally prefers pre-verbal elements to mark the experiential (Yuan et al. Reference Yuan1989), this preference cannot be assumed for Hokkien. According to Bodman (Reference Bodman1955), the Hokkien experiential aspect may be marked post-verbally by -ke (cognate of Mandarin -guo, Cantonese -go), pre-verbally by bat or by the combination bat … -ke. The three sentences in (2), cited from Bodman (Reference Bodman1955: 269–270), are interchangeable (cl = classifier; asp=aspectual marker):

1. (2)

   1. (a) cit e sian-si i-cheng cue-ke kuaː

   2. (b) cit e sian-si i-cheng bat cue kuaː

   3. (c) cit e sian-si i-cheng bat cue-ke kuaː

      this clgentleman before asp be-asp official

      ‘This gentleman used to be an official.’

The same variation is found in Singaporean Hokkien (Zhou & Chew Reference Zhou and Chew2000). Hokkien, Cantonese, and Mandarin are convergent in the way the experiential aspect is expressed. In terms of position, (2b) is congruent, and (2c) is partially congruent, with English ever. We need to assume that bat is dominant over -ke or bat … -ke in Hokkien, an assumption for which we have no empirical evidence. For the construction-based model (1b), the three-way marking of the experiential in Hokkien does not affect transferability. As the exponent of the Chinese-derived experiential aspect, the position of ever in Singapore English is determined by its position in English, regardless of the position of bat in Southern Min or cenjing in Mandarin.

The behavior of the imperfective poses a more serious challenge to the congruence model. Across Chinese dialects, the imperfective may be used with agentive and stative predicates. In Chinese linguistics, the former is often called progressive, and the latter, continuous (see Chappell Reference Chappell1992). The progressive is marked by a pre-verbal morpheme (Mandarin zai, Hokkien teq) and the continuous, by a post-verbal particle (Mandarin -zhe, Hokkien -teq). The two imperfectives in Hokkien, adapted from Bodman (Reference Bodman1955), are illustrated in (3).

1. (3)

   1. (a) Progressive: i teq thak cheq (Bodman Reference Bodman1955: 151)he asp read book‘He is studying.’

   2. (b) Continuous: i tou-teq khuaː cheq (Bodman Reference Bodman1955: 325)he lie-asp see book‘He reads while lying down.’

The continuous use of -teq provides the background or circumstance to the event expressed by the main verb. As with the experiential, we see clear convergence among Hokkien, Cantonese and Mandarin in the imperfective.

Siegel remarks that the English word keep is felicitous with the continuous (e.g. keep running), and due to the congruence constraint, the continuous function is not transferred from the post-verbal -teq to the pre-verbal keep in Singapore English. There is, however, congruence between teq and keep, and between -teq and -ing. Still, there is no transfer. The model in (1b) offers a different perspective on the issue. According to Bao (Reference Bao2005), the Chinese imperfective transfers to Singapore English, together with the other aspectual categories in the Chinese system, and is exponenced by the English suffix -ing, which is the most felicitous morpheme as the exponent of the Chinese imperfective. As in English, there is no explicit progressive–continuous distinction in Singapore English. The failure of the continuous to transfer is not due to the lack of a suitable target, but to the general requirement of English grammar (*He lying-down reads). It is filtered out of the Singapore English system. The difference in morphosyntactic position does not determine the exponencing of the transferred grammatical construction.

In conclusion, I would like to draw readers' attention to the debate among generative phonologists on the phonetics and phonology of lexical borrowings. There are two opposing camps. One camp argues that loanwords are adapted on the basis of phonetic cues, while the other camp maintains that the sound systems of the source and borrowing languages play a decisive role (see Kang Reference Kang, van Oostendorp, Ewen, Hume and Rice2011). The two competing models in (1) parallel the phonetics–phonology divide in the analysis of loanword adaptation, and offer different perspectives on, and analyses of, grammatical borrowings. For Siegel (Reference Siegel1999, Reference Siegel2008), surface order is crucial; for Bao (Reference Bao2005, Reference Bao2009), it is the grammatical system that drives substratum transfer.