The Periphrastic Topic Structures in Chinese-English Interlanguage

Periphrastic topic structures, as Chinese-style topic structures, belong to the category of prepositional topic fronting constructions in TSVO sequences. Findings from studies on periphrastic topic structures are inconsistent and present only a fragmented understanding. Therefore, the present study is conducted to make up for the gap and aims to reveal the developmental features of periphrastic topic structures (henceforth PTS) through investigating the production and the recognition of PTSs in Chinese college English learners’ Chinese-English interlanguage. The result suggests that with advances in learners’ English proficiency levels, periphrastic topic structures diminish from the preliminary stage to the intermediate stage gradually, but present a much higher degree of fossilization at the advanced level. Theoretically, this finding may further support Yang’s findings(2008) and validate the Selective Fossilization Hypothesis model (SFH model) proposed by Han (2009).


INTRODUCTION
Chinese is claimed to be a topic-prominent language (henceforth TPL), while English is categorized into a subject-prominent language (henceforth SPL) (Li and Thompson, 1976;Xu and Langendoen, 1985). The subject performs a crucial function in the subject-predicate constructional sequence while the topic is an indispensable element in the topic-comment syntactic structure. Chinese English learners, due to their deficient mastery of target language pragmatic objectives, often produce a large quantity of erroneous English sentences possessing Chinese grammatical characteristics at different stages of SLA (Yang, 2008). Yang (2008) studied six types of topic-prominent constructions produced by three groups in a spontaneous oral task and a careful translation task including Double Nominative (DN), Existential Constructions(EN), Pseudo Passives(PP), Null Element(NL), Periphrastic Structures(PTS), and Subject-predicate Disagreement (SPD) and found that with the promotion of learners' English proficiency levels, some topic-prominent parameters would constantly decrease, but others, especially PTS, would plateau even at advanced stages of TL development. Based on Yang's findings, the current study further aims to explore the developmental features of periphrastic topic structures in the Chinese-English interlanguage (henceforth CEIL) of Chinese college English learners among different proficiency levels so as to draw conclusions concerning whether or not cross-linguistic typological transfer interacts with the selective fossilization during the process of second language acquisition. The research target of the investigation is periphrastic topic structures which belong to Chinese-style topic-prominent syntactic constructions and exist pervasively in Chinese English learners' CEIL, even at the advanced proficiency level (Yang, 2008).Therefore, on the basis of philosophies about the cross-linguistic typological transfer and the Selective Fossilization Hypothesis (Han, 2009), two principal research questions were formulated in an effort to deepen our understanding about periphrastic topic structures in the Chinese-English interlanguage: (1) Does learners' topic-prominent syntactic structure knowledge interfere with their production and recognition of periphrastic topic structures in a Chinese-English translation test and a grammaticality acceptability judgment test? (2) What is the developmental pattern of the periphrastic topic structures in college English learners' Chinese-English interanguage? It is expected that the current study will reveal the developmental patterns of Chinese English learners' Chinese-English interlanguage from the perspective of periphrastic topic structures and provide support for the Selective Fossilization Hypothesis Model (Han, 2009). What's more, the analysis will provide some evidence for investigating the IJALEL 10(5):57-67 development of interlanguage in specific aspects with the theoretical framework of fossilization.

Typological Language Transfer Philosophies
For speakers of a TPL learning a SPL, typological language transfer occurs when learners apply TPL properties into SPL syntactic devices so that the process of second language acquisition is promoted or suspended. In respect to typological transfer in the field of SLA, there exist two mutually opposite controversies (Yang, 2015). One view holds that interlanguage at the early stage of SLA possesses the universal feature of "topic-comment" and is irrelevant to the learners' native language (henceforth NL). Odlin (1989), Klein (1986), Clahsen and Muysken (1986) asserted that English learners from different L1 backgrounds produced similar topic-comment structures, which validated that there was no typological transfer in interlanguage systems. The other view states that the topic-prominent stage in the initial stages of learning is evidence of language typology transforming from L1 to L2. Schachter (1979) and Rutherford (1983) studied English sentences produced by learners whose native language was TPL, which demonstrated that English learners from TPL produced a large quantity of English sentences possessing obvious features of TPL. Furthermore, Sasaki (1990) investigated the existential constructions in the written language produced by Japanese English learners (with Japanese being a TPL) and claimed that with advances in students' proficiency levels, the use of TPL syntax was transformed gradually into the more appropriate use of SPL syntax. Along these lines, Givón (1995) argued that the initial stage in SLA was parallel to TPL, which meant there was a systematic transferable process from L1 to L2 rather than the occurrence of a universal TPL level. Finally, after noting a lack of research regarding possible transfer for speakers of SPLs learning TPLs, Jung (2004) studied English learners' acquisition of Korean (a TPL) and validated that learners applied SPL syntax in their TPL constructions.

The Selective Fossilization Hypothesis Model
Han (2009) first proposed the Selective Fossilization Hypothesis (henceforth SFH) in interlanguage so as to extend the field of fossilization research. SFH considers two influential factors which engender selective fossilization, including L1 markedness and L2 input robustness. Han suggests that each of these is determined by two sub-variables: Frequency and variability. Frequency represents the quantitative properties of a given linguistic usage in the L1 or the L2. Variability refers to the form-meaning function of a given linguistic usage (Han, 2013).
According to Han (2009), the given usage in one's TL occurs on an input robust continuum between ROBUST (frequent, invariable) and NON-ROBUST (infrequent, variable). The corresponding usage in one's NL appears on a markedness continuum between MARKED (infrequent, variable) and UNMARKED (frequent, invariable) ( Figure 1). Each parameter is determined by the interaction and the relevant strength of sub-variables of frequency and variability (Qi, 2009). Figure 1 explicates the intersectional functions about L1 markedness and L2 input robustness in order to demarcate four zones indicating four conditions in SLA. The first zone stands for UNMARKED in L1 and ROBUST in L2; the second zone signifies MARKED in L1 and ROBUST in L2; the third zone symbolizes MARKED in L1 and NON-ROBUST in L2, the fourth zone embodies UNMARKED in L1 and NON-ROBUST in L2. Accordingly, the first and third zones (I and III) are defined as either the acquisition zone or the fossilization zone respectively, depending on a variety of factors. The second and fourth zones (II and IV), however, represent the acquisition and fossilization zones respectively. Han(2009) argues that structures that fall into zone II are likely to be acquired by L2 learners, while those falling into zone IV are likely to fossilize before NL-like mastery is achieved. The concentric circles from inner to outer symbolize various degrees of acquisition or fossilization. The most outside circle in zone IV, for example, represents the highest degree of fossilization.

Definition and features
Periphrastic topic structures originally derive from the marked topic constructions induced by prepositional phrases in Chinese. They belong to a category of "Chinese-style" topic structures, a term originally advanced by Chafe (1976). The origin of "periphrasis" is a Greek word in which "peri" means "round about" and "phrase" means "to express". Thus, the periphrastic structure is used to express something in a round-about way. This structure is what Yip and Matthews (1995) term "periphrastic topic constructions". They claim that the periphrastic topic construction is a common interlanguage structure which represents topic-prominent features in interlanguage. The Chinese syntactic structures contain prepositional phrases like 对于(duiyu), 关于(guanyu), 至于(zhiyu), and 对于…来说(duiyu…laishuo) which are equivalent to 'for…', 'as for…', 'about…', or 'speaking of…'in interlanguage. The periphrastic structure, often followed by a noun, a phrase, a sentence and sometimes with an intonation break, sets a boundary between topic and comment construction (Yang, 2008: 126).
There are many semantic and grammatical constraints in English topic constructions. The co-referential relationship occurs between the topic and some constituent or a null element in the comment. In Chinese, the semantic constraints in topic structures are relatively less and the relation between the topic and the comment is loose. Chinese English learners transfer the loose topic-prominent structures in Chinese into the process of producing English structures. Therefore, the phenomenon is defined as a new kind of transfer process that results in expressing NL topic-prominent functions through TL syntactic forms (as seen in [1]).
(b) For successful men, people pay more attention to their social status instead of their capability. The syntactic structures of (a) and (b) are the same, but there exist a few differences in terms of the topic structures. The relationship between the topic and the comment is relatively loose in (a) whereas the relationship between the topic and the comment in (b) is closely correlated. In (a), the constituent chenggongnanxing ('successful men') in the topic and the constituent tamende ('their') in the comment may not refer to the same kind of people. Therefore, the relation between the topic and the comment in (a) is universally characterized as "aboutness" in the literature (Yuan, 2017: 44). In (b), the constituent 'successful men' in the marked topic structure is co-referential with the pronominal element 'their' in the comment construction.

Classifications
In periphrastic topic constructions, three dimensions of relationships are identified between topics marked by prepositional phrases and their comments, including a syntactic relationship, a semantic relationship, and a pragmatic relationship (Yuan, 1996;Liu, 1998/2003;Nie, 2007;Zhou, 2012). Accordingly, periphrastic topic structures in CEIL are divided into the following five subcategories.
In terms of the syntactic dimension, there exists the argument co-indexing relationship (henceforth AC) between the topic and subject or object in the comment. That is to say, the subject or the object connected by the predicate in the comment component constitutes the co-indexation with the topic by means of null elements, pronominal elements, and co-occurrence with the nominal phrase (henceforth NP) in the topic (Yuan, 1996;Xu and Liu, 1998), as seen in (2). (Note that the English represents literal translations that retain the Chinese word order.) 'As for this question, teacher Zhang will explain it better than me. ' (see Appendix A) The prepositional phrases guanyu and duiyu in (a) and (b) belong to the category of topic markers. The subject of the comment in (a) forms an anaphora with the topic zhegetiyi ('this proposal') by means of null elements. In (b), the object in the comment is co-indexed with the topic zhegewenti ('this question') through null elements. (The square brackets in two sentences stand for null elements.) In terms of the semantic dimension, firstly, there is the valence relationship (henceforth VR) between the topic and the NP in the comment. In Chinese, NPs possess the design feature of valence (Yuan, 1992). The valence of NPs refers to the phenomenon that the valence NP and its related NP comprise a semantic dependent relationship. In periphrastic topic structures, the topic marked by prepositional phrases is a sememe (downgraded object or downgraded argument or semantic component) of the valence NP in the comment (Yuan, 1992;Herbst, 1988;Zhu, 1998). In other words, the valence NP in the comment and the NP in the topic form an integrated logical semantic relationship (Xu and Liu, 1998) (as seen in [3] In (a), the topic constituent lixiang ('the dream') is a sememe of the valence NP xiangfa ('idea') in the comment. Therefore, there exists a valence relationship between the topic and the NP in the comment. The object yijian ('opinions') in the comment of (b) is the valence NP and the topic is a semantic component of it. The possessor of yijian ('opinions') refers to me ('the speaker') and focuses on the work arrangement.
There still exists the superordination-hyponymy register relationship (henceforth SH) between a topic and an argument connected by the predicate in a comment (Xu and Liu, 2003;Yuan, 2017). Specifically, the topic is the superordinate concept for the argument in the comment. Both constituents express a universal set and subset respectively, such as in fruit and apple, or animal and tiger (as seen in [4] In (a) and (b), the topics Beijingdemingshengguji ('the historical sites in Beijing') and zhongguocai ('Chinese dishes') on the one hand, and the arguments yiheyuan ('the Summer Palace') and Beijingkaoya ('Beijing Roast Duck') in the comment on the other hand, form the hyponymy register relation with the topic being the superordinate and the NP in the comment being its hyponym. Furthermore, the hyponymy conception, which is connected by predicative verbs in syntactic structures, is transformed into the following judgment propositions.
In terms of the pragmatic dimension, there are the background register topics incorporated into these structures, where the relation between the topic and the comment is the loosest, and the relevant syntactic-semantic connection does not appear between a topic and a nominal constituent in a comment. The established relationship between a topic and a comment is universally characterized as "aboutness" in the literature, which is dependent on background knowledge and discourse context (Chao, 1968;Shi, 1998;Liu, 1998/2003;Nie, 2007;Yuan, 2017). These constructions, in which topics do not possess any obvious syntactic and semantic relationships with comments, are classified into this type, including two subcategories. In the first subcategory, there exists a definitive judgment relationship (henceforth DJ) in which the topic constrains the judgment degree of the comment, and the content of the comment is dependent on the topic (Xu and Liu, 2003) (as seen in [5] The topics in (a) and (b) define the semantic domains of the comments. If the prepositional constructions inducing the topics are deleted, the semantic relation between the topics and the comments will be contradictory.
There is still a theme-content relationship (henceforth TC) between the topic and the comment. In other words, the topic functions as the theme of the comment and the comment is the further explanation for the topic (Nie, 2007) (as seen in [6]).
Guanyu xianzhi sijiacheshulian, ta daibiao zhichide yifang. 'As to putting a limit on private cars, he is in support of it.' (b) 对于感情，我只知道我们俩相互喜欢。 Duiyu ganqing, wo zhizhidao womenlia xianghuxihuan. 'As for emotion, I only know that we fall in love with each other. ' (see Appendix A) In (a), there exist no semantic relationships between the topic xianzhisijiacheshuliang ('putting a limit on private cars') and the syntactic constituents ta ('he'), daibiao ('stands for'), and zhichideyifang ('in support of it') in the comment. Similarly, in (b), the content incorporated into the comment is dependent on the topic ganqing ('the emotion').
In a word, the comments provide further explanation and interpretation for the topics. This semantic relation, therefore, is beyond the syntactic category.

Related Studies about TPL and SPL
Some studies have concentrated on the typological influence of topic-prominent structures in the NL during the process of acquiring English as a target language (henceforth TL). Fuller and Gundel (1987), for example, examined the functions of topic-prominent structures in the process of adults' acquiring English as a second language. They found support for the hypothesis that the general TPL features could influence the developmental route of interlanguage. Yip (1995) investigated the design features of two interlanguage syntactic structures, namely, Pseudopassives and periphrastic topic structures. The subjects were intermediate and advanced English learners from Taiwan, Mainland China and HK. He came to the conclusion that Chinese English learners often utilized English syntactic structures to perform topic communicative functions in Chinese. In addition, Yang(2008) studied six types of topic-prominent constructions produced by Chinese English students among three groups in a spontaneous oral task and a careful translation task and found that learners among three proficiency groups, to some extent, have transferred the topic-prominent structures to the Chinese-English interlanguage and they tend to produce more subject-prominent constructions with the development of their English proficiency levels.
On the contrary, other studies have delved into the cross-linguistic role of subject-prominent structures with English native speakers as subjects and a TPL as the acquired entity. Jung's investigation (2004) contradicted the hypothesis that there existed an early general stage of topic-prominence in L2 acquisition. With the promotion of learners' L2 performance, they gradually became more conscious of the normal structures of the TL. Furthermore, Liu (2015) examined the developmental process for English native speakers' acquiring topic-prominent sentences in Mandarin Chinese. His results indicated that low-level subjects behaved on a par with native speakers for certain types of topic structures, and these learners could produce new constructions at the initial stage of learning. Similarly, Yuan (2017) investigated English native speakers' acquisition of Chinese base-generated-topic sentences. The results showed that English native speakers were more likely to make full use of TL syntactic devices and semantic information, rather than NL devices and information, during their processing of Chinese base-generated-topic sentences.
There exist some limitations about these previous studies. For one thing, the majority of investigations have investigated the developmental patterns of TPL and SPL during the process of SLA in a broad sense, but few studies touch the specific aspects of interlanguage development. Furthermore, most relevant research has explored the interference of cross-linguistic typological transfer between NL and TL, but ignored the influence of fossilization during the different stages of SLA.

Participants
Ninety Chinese English learners at a prestigious university in China were selected as the participants. They were then divided into three groups representing different English proficiency levels. Group One represented the preliminary level and these students consisted of freshman who were non-English majors. Group Two was the intermediate level and they were English major junior students. Group Three was the advanced level with third-year postgraduate students majoring in English linguistics. Subjects in the preliminary group had obtained excellent English grades in their entrance examinations and learned English for at least eight years. The full mark of the college entrance English examination in China is 150 and a score range of 120 to 150 stands for the excellent level. Participants in the intermediate group had passed the TEM-4 (with a score range of 70 to 79) and learned English for almost eleven years. Subjects in the advanced group had passed the TEM-8 (with a score range of 70 to 79) and learned English for at least fifteen years. The TEM-4(Test for English Majors-Band Four) and the TEM-8(Test for English Majors-Band Eight) are the two standardized tests about English proficiency administered by The National Advisory Committee on Teaching English Language to Majors in Higher Education under the Ministry of Education of the People's Republic of China in an effort to measure college English majors' comprehensive English performance including listening, speaking, reading, writing, and translation. The TEM-8 is more advanced than the TEM-4. The full marks of the two tests are 100 and the scores of them have the same ranks. A score range of 60 to 69 stands for the passing marks, 70 to 79 represents the relatively good marks, and 80 to 100 signifies the excellent marks. Additionally, the dividing boundary is not absolute.

Instruments
Two instruments were used in the current research: a Chinese-English translation test (CET) and a grammaticality acceptability judgment test (GAJ). The Chinese-English translation test, as a productive task, aimed to investigate whether the subjects among the three groups transferred the topic-prominent structures in Chinese into their CEIL directly. The grammaticality acceptability judgement test, as a recognition task, aimed to examine whether the subjects in the three groups had completely acquired the subject-prominent syntactic knowledge in English.
The Chinese-English translation test was comprised of twenty Chinese sentences. The stimuli relevant to the study was the five types of periphrastic topic structures including AC, VR, SH, DJ, and TC, with each type in three test sentences (see Appendix A for all the 15 test sentences used in these 5 types). The other five sentences were distractor sentences. They were randomly embedded among the fifteen test items. This test was designed to elicit subjects' production of periphrastic topic structures. Considering the fact that learners' individual L2 performance may exert a varying influence on their production, only high-frequency words were included in the test sentences of the two tasks.
The grammaticality acceptability judgment test consisted of twenty English sentences. Again, there were three sentences representing each type of periphrastic topic structures and five additional distractor sentences (see Appendix B for all the 15 test sentences used in these 5 types). The test sentences were presented in a random order. This task was analogous with the one designed by Yuan (2017) in order to examine learners' awareness and sensitivity to the subject-prominent syntactic structures in English. Participants were instructed to judge the degree of grammatical acceptability for the target sentences using a four-point Likert scale. The four options were transformed into 1, 2, 3, and 4 respectively. 1 represented Completely Unacceptable. 2 signified Probably Unacceptable. 3 represented Probably Acceptable. 4 indicated Completely Acceptable. 2 and 3 were included in the Likert scale in order to capture subjects' simultaneous recognition for the target sentences (also see the instructions in Appendix B).

Procedures of Data Collection and Analyses
Data collection and analyses were carried out with the aid of EXCEL and SPSS 24.0. First, the five types of periphrastic topic structures in the two tests were manually marked with AC, VR, SH, DJ, and TC respectively. Next, the relative frequencies of produced periphrastic topic structures among the three proficiency levels in the Chinese-English translation test and the inaccuracies for recognizing the target sentences in the grammaticality acceptability test were counted and tabulated into a table using EXCEL. Then, the comparisons about the numbers and the percentages of productions in the CET and inaccuracies of recognition in the GAJ were made so as to reveal the developmental tendency of periphrastic topic structures. Finally, the production and the inaccuracy of recognition about each PTS parameter in the two tests were investigated using one-way ANOVA in SPSS 24.0 in an effort to reveal the significant difference between each group.

Cross-linguistic Typological Transfer from NL to TL
The standards to judge subjects' production in the Chinese-English translation test depended on whether the sentences were grammatically correct in English and conformed to the subject-prominent syntactic structures. The target sentences in the grammaticality acceptability judgment test originated from the literal translation of prepositional topic fronting constructions in Chinese. Therefore, numbers three (Probably Acceptable) and four (Completely Acceptable) represented faulty choices.
The numbers and relative frequencies of produced PTSs in the CET and the inaccuracies for recognizing PTSs in the GAJ are shown in Table 1. Table 1 reveals that the inaccuracies for recognizing periphrastic topic structures in the GAJ at each proficiency level are larger than the relative frequencies of produced PTSs in the CET (82.67%>73.11% at the preliminary level; 57.33%>54.67% at the intermediate level; 63.56%>57.33% at the advanced level). In the CET, when learners at the IJALEL 10 (5):57-67 intermediate and advanced stages produced TL structures after they had read the Chinese constructions, they may have resorted to their previously acquired TL knowledge and utilized processing strategies to check whether their produced syntactic structures accorded with the expressive norms of the TL. The fifteen English periphrastic sentences provided in the GAJ, however, correspond with prepositional topic fronting constructions in Chinese. Subjects among the three groups probably assumed that those sentences conform with semantic and pragmatic norms of the TL in their potential consciousness rather than employing their multi-dimensional processing mechanisms to test the applicability of such constructions in English.
Therefore, a general conclusion can be drawn that subjects tend to rely almost exclusively on topic-prominent structures in Chinese to produce and recognize periphrastic topic structures in the two tests, which answers the first research question. In a word, the cross-linguistic typological transfer from the NL permeates the whole continuum of SLA. This finding does not validate Odlin's (1989), Klein's (1986), Clahsen & Muysken's (1986) hypothesis that there was no typological transfer in interlanguage systems, but reveals that some salient linguistic items from NL was transferable to the production of TL during the process of Second Language Acquisition (Schachter, 1979;Rutherford, 1983;Sasaki, 1990;Givon, 1995;Yip, 1995;Jung, 2004;Yang, 2008).
Periphrastic topic structures are the most common in prepositional topic fronting constructions in Chinese. The finding supports the notion that the conceptual system of the NL, and learners' previously acquired incomplete competence of the TL, can jointly affect the learners' CEIL in respect to these structures and their comparative absence in the TL (see also Yang, 2008). In Chinese, the prepositional topic fronting structure is classified into an unmarked construction. Contrastively, the corresponding structure in English is unusual and is defined as a marked structure. The markedness theory advanced by Hyltenstam (1984) held that the corresponding unmarked structure in the NL was unconsciously transformed into the usage of TL when the construction in the TL was a marked one. In general, this type of transfer is unconscious and is defined as Transfer to Nowhere (Andersen, 1983).

The Developmental Feature of Periphrastic Topic Structures
The multiple comparisons for each PTS type between each proficiency level are conducted so as to reveal the developmental tendency of periphrastic topic structures and the statistical results are shown in the following tables.
According to Table 2, in the CET, none of the multiple comparisons revealed significant differences between three groups. In the GAJ, the difference between the intermediate The difference between the preliminary group and the intermediate group, and that between the preliminary group and the advanced group, however, showed the statistical significance (p = 0.003; p <.001).
As seen in Table 3, in the CET, there was statistically significant difference between the preliminary group and the advanced group (p = 0.041). However, the difference between the preliminary group and the intermediate group and that between the intermediate group and the advanced group exhibited non-significance (p = 0.437). In the GAJ, the difference between the preliminary group and the intermediate group and that between the preliminary group and the advanced group were significant (p = 0.001; p = 0.045). In addition, the difference between the intermediate group and the advanced group was not significant (p = 0.407).
According to Table 4, in the CET, the difference between the preliminary group and the intermediate group and that between the preliminary group and the advanced group showed the statistical significance (p = 0.011; p = 0.027). The difference between the intermediate group and the advanced group showed non-significance (p = 0.942). In the GAJ, the difference between the preliminary group and the intermediate group and that between the preliminary  group and the advanced group were statistically significant (p <.001; p = 0.011). However, the difference between the intermediate group and the advanced group was not significant (p = 0.123). As shown in Table 5, in the CET, the difference between the preliminary group and the advanced group and that between the intermediate group and the advanced group were not significant (p = 0.645; p = 0.179). The difference between the preliminary group and the intermediate group was significant (p = 0.023). In the GAJ, none of the multiple comparisons revealed significant differences between the groups.
According to Table 6, in the CET, the difference between the preliminary group and the intermediate group was significant (p = 0.019). The difference between the preliminary group and the advanced group and that between the intermediate group and the advanced group were not significant (p = 0.281; p = 0.441). In the GAJ, none of the multiple comparisons indicated significant differences between the groups.
As mentioned above, the difference between the preliminary level and the intermediate level showed the significant difference. However, the difference between the intermediate level and the advanced level displayed statistical non-significance. It is demonstrated that periphrastic topic structures diminish from the preliminary stage to the intermediate stage gradually, but present a much higher degree of fossilization at the advanced level, which explains the second research question.  According to the SFH model (Han, 2009), the TSVO sequence, as a Chinese-style topic structure, is more frequent, less marked, and less variable in Chinese than in English. Thus, it falls on the unmarked category on the L1 markedness continuum. In contrast, TSVO is a peripheral syntactic structure in English, which is less frequent, more marked, and more variable in English than in Chinese. Therefore, it is labeled as NON-ROBUST on the L2 input robustness continuum. Accordingly, the periphrastic topic structures in CEIL are defined as the UNMARKED(frequent, invariable) on the Chinese markedness continuum and are labeled as the NON-ROBUST (infrequent, variable) on the English input robustness continuum, which reveals that periphrastic topic structures fall into the fourth zone(the fossilization zone). In addition, the fossilization of periphrastic topic structures in CEIL can be further expounded by the Numerical Value Model of SFH (Han, 2009). The two paramount variables of SFH (L1 markedness and L2 input robustness are exemplified by means of the following formula: P foss =A/M 11 -BR 12 (A>0; B>0) In this formula, P foss stands for the probability of fossilization. M 11 symbolizes L1 markedness. R 12 signifies the input robustness of L2. The signs A and B refer to coefficients. M 11 leads to fossilization while R 12 is inversely proportional to the appearance of fossilization. In other words, the probability of fossilization equates UNMARKED of L1 and NON-ROBUST of L2. The periphrastic structure is labeled as UNMARKED in Chinese and NON-ROBUST in English. Accordingly, the fossilization probability of periphrastic topic structures is much higher for Chinese learners of English than for English learners of Chinese.

CONCLUSIONS
The current study mainly explores the developmental features of periphrastic topic structures in the Chinese English learners' CEIL. Two conclusions can be drawn on the basis of data analysis, results and discussions enumerated above. First, periphrastic topic structures, as Chinese-style topic-prominent constructions, are transferable to the production of Chinese college English learners' Chinese-English interlanguage, which has provided the evidence that cross-linguistic typological transfer appears pervasively in the process of second language acquisition. Second, the periphrastic topic structures in Chinese college English learners' CEIL tend to diminish from the preliminary stage to the intermediate stage gradually but present a much higher degree of fossilization at the advanced level. Therefore, it is universally concluded from the two findings that the cross-linguistic typological transfer from NL and the selective fossilization of some NL linguistic items pervade the entire process of second language acquisition, which further validates Yang's findings (2008) and confirms the Selective Fossilization Hypothesis model (SFH model) proposed by Han (2009).