Acquisition of Double-Nominative Constructions by Italian L1 Learners of Chinese A Cross-Sectional Corpus Study

This paper presents new results of an ongoing cross-sectional corpus study investigating the acquisition of Chinese word order by Italian L1 learners. Specifically, it focuses on the acquisition of ‘double-nominative constructions’, as well as the correct sequential organisation of topical and focal information in the Chinese sentence. The analysis is conducted on three learner corpora, created by the Author on the basis of a test submitted to three groups of university (BA and MA)-level Italian L1 learners of Chinese, for a total of 132 learners. Quantitative and qualitative analysis conducted on the collected data show that, while the double-subject construction may appear as a simple and straightforward pattern, it is in fact a rather difficult construction to acquire and spontaneously produce for Italian L1 learners. Rather, students tend to use patterns they are used to in their L1 (or other L2s, such as English). These include the [NP1 have NP2], [NP1 的 NP2], or [NP1 adjectival predicate ] patterns, among other types, thus confirming the inhibitive L1 transfer hypotheses of this study.


Introduction
This paper presents new results of an ongoing cross-sectional study investigating the acquisition of Chinese 1 word order by Italian L1 learners.The focus of this paper is the acquisition of so-called "double-nominative constructions" (Chappell 1996;Shen 1987;Modini 1981;Li, Thompson 1976, 1981;Tsao 1977;Mullie 1933;Teng 1974b;Li 2004, see section 3) as well as the correct sequential organization of topical and focal information in the Chinese sentence.Double nominatives 2 generally refer to a pattern consisting of two nominals that occur next to each other at the beginning of the sentence, followed by a predicative element (henceforth, PE), e.g., an attributive or stative verb, as in (1.C).Such constructions are tightly connected with the topic-prominent nature of the language, which is a fundamental typological feature of Chinese: as discussed in detail in section 3, this construction allows hosting topical elements (NP1, dàxiàng 'elephants') in the sentence-initial position, while maintaining focal (informationally salient) elements at the very end (in this case, cháng 'long' or bízi cháng 'nose (is) long' depending on the context).However, such a pattern is not allowed in most Indo-European languages of Europe, including English (1.E) and Italian (1.I).Rather, such languages may, for example, use the verb have to express that possession/whole-part relationship, and specifically the [NP1 have NP2] pattern, as in (2.E) and (2.I').Thus, students are not familiar with the double-nominative pattern and may find it difficult to spontaneously produce it.Rather, they may tend to produce patterns that are closer to the structures 1 In this paper, I use the term 'Chinese' to refer to Putonghua, the standard language of the PRC.Simplified Chinese characters and the Pinyin romanization system have been used throughout the article.The glosses follow the general guidelines of the Leipzig Glossing Rules.A list of all abbreviations and glosses is available at the end of the article.
2 The term 'double nominative' is here used for the reader's convenience, i.e., to allow readers to refer to the relevant literature on the topic.It is nevertheless important to note that grammatical cases (intended as a system of marking of a dependent nominal for the type of relationship they bear to their heads, such as nominative, accusative, ablative etc. in Latin or German) are not part of the grammar of Chinese (see section 3).
of their native language, e.g., the [NP1 have NP2] pattern as in (2.C), which however is a less preferred pattern in Chinese in terms of information structure 3 (section 3): This paper specifically looks at negative L1 transfer phenomena that affect the production of double-nominative constructions by Italian L1 learners of Chinese.Specifically, it presents the results of an ongoing cross-sectional study consisting of translation tasks submitted to three cohorts of Italian L1 learners of Chinese with different proficiency levels and from different universities, for a total of 132 learners.The test has been elaborated on the basis of existing crosssectional studies conducted on English L1 learners of Chinese (Jiang 2009) and on a preliminary contrastive analysis of the form-function associations connected to this construction in Chinese and Italian.The analysis singles out different linguistic forms and patterns used by Italian and Chinese to encode the same meaning, which may lead to negative transfer and high error rates.The data collected through the test are then used to determine whether there is a correlation between L1-L2 differences, the proficiency level of the students, the complexity of the tested structure, and the percentage of the related word order errors in L2 production due to L1 transfer.This work aims to contribute to Chinese as a Second/Foreign Language (CSL/CFL) acquisition as a growing area of research.The past decade has witnessed an increasing interest in CSL/CFL acquisition.However, most studies have been conducted among English L1 learners, while there exist very few studies on Italian L1 learners of Chinese, especially on word order acquisition.This study wants to contribute filling this gap.An increasing number of Italian educational institutions are now offering Chinese courses, not only at the university level, but also in primary and secondary schools.The findings of the study will contribute to a clearer understanding of the acquisitional processes of double nominatives by Italian L1 learners, thus building up towards a more effective pedagogy.
3 This can be observed in corpora, as well.The string "鼻子很长" bízi hěn cháng occurs 11 times in the the BCC corpus of Modern Chinese (Beijing Language and Culture University, 15 billion characters), while no occurrence can be found of the string 有很 长的鼻子 (Xun 2019).Obviously, the use of one over the other form depends on the context and on the information structure/focus of the sentence.

382
The rest of the paper is organised as follows: section 2 presents the approach the study adopts, namely that of form-to-function studies (Long, Sato 1984;Bardovi-Harlig 2014, 2015), along with its basic notions, tools, and research methods.Section 3 offers a description of double nominative constructions, their formal characteristics and their functions, as well as a contrastive analysis with Italian and English.Section 4 describes the study, and in particular the hypotheses, the test design, the participants, and information on data collection.Section 5 presents the results, both from a quantitative and a qualitative perspective.Section 6 discusses these results, while section 7 draws the conclusions and proposes some indications for CSL/CFL teaching.

2
The Approach This study adopts a functional approach to the investigation of second language acquisition (henceforth SLA), and specifically that of form-to-function studies (Long, Sato 1984;Bardovi-Harlig 2014, 2015) in line with previous studies on related issues (Morbiato 2017).Such an approach is particularly useful for several reasons.First, a formoriented (form-to-function) approach allows a comprehensive investigation of a specific construction/aspect (form) of the L2 (e.g., the English past tense -ed, or in this case the double-nominative construction) by examining its different facets: it investigates its use in language to identify its functions (which may be multiple) and compares these functions to correspondent forms in students' L1; then, it verifies whether and to what extent learners use the target form in L2 production.Second, it provides useful tools and notions that allow analyzing the complex process of acquisition of the target L2 form, including: the importance attributed to learners' errors, that are not seen as students' failures, but rather as SLA diagnostic tools allowing investigation of acquisitional processes at different stages (section 2.1); the role of interlanguages, i.e., idiolects developed by each L2 learner, which evolve over time and include errors (section 2.1); the notion of 'linguistic transfer' and, consequently, the crucial role attributed to contrastive analyses of L1-L2 differences to predict transfer-related acquisitional issues (section 2.2).Third, it encourages to examine the 'inventory' of means (forms) learners use to express a specific meaning (function), as well as and the 'reorganization of the balance' 4 of these means over time and across dif- ferent proficiency levels.This is based on the tenet that learners already have access to a full range of concepts (functions), as well as the means (forms) to express these concepts in their L1, but lack the related forms in their L2, which may significantly differ from forms in their L1.An awareness of students' meaning-form associations is crucial for language pedagogy: recent researches have shown that a focus both on form and on meaning/communication can have advantages over purely meaning-focussed instruction (Alcón-Soler 2012).
The next subsections present in greater detail concepts mentioned above that are important for the present analysis (for a more detailed discussion of the framework see Bardovi-Harlig 2015).

L2 Errors and Interlanguage
While students' L2 errors used to be seen as something to be eradicated, after Corder's article (1967) and subsequent literature on the topic, they started to be considered as significant diagnostics to monitor learners' acquisitional processes.Now, in SLA research, L2 errors are regarded as part of the strategies each learner develops in the process of creating their own interlanguage."Interlanguage" (Selinker 1972(Selinker , 2012) ) is seen as a dynamic and evolving linguistic system with its own code and rules, and with an unstable nature.It is mostly explained as the result of an interaction of two language systems, usually the mother tongue and the target language (TL).The analysis of interlanguage development, of common errors and error types, and of the reasons beyond such errors, leads to a clearer understanding of the learning stages and the learners' difficulties caused by L1 interference (see section 2.2).Errors refer not only to forms that are grammatically incorrect, but also forms that are semantically or pragmatically inappropriate in the context: an L2 error is also a linguistic form […] which, in the same context […] would in all likelihood not be produced by the learner's native speaker counterparts.(Jiang 2009, 126) An awareness of how errors (and error types) evolve over time or across different proficiency levels contributes towards more effective pedagogical actions that help learners self-monitor, cope with, and eventually overcome persistent errors more efficiently and effectively.
'then').The following stage involves the use of tense, verbal morphology, and so on.In this process, the inventory of forms changes (i.e., it increases over time) and the balance changes as well, as the use of morphology overrides temporal sequence, e.g., in sentences such as 'I entered after buying the ticket', which display reverse chronological order.

Cross-Linguistic Influence and Linguistic Transfer
The term 'transfer' in general refers to "the extent to which knowledge and skills acquired in one situation affect a person's learning or performance in a subsequent situation" (Ormrod 2014, 206).Transfer affects all areas of language; it is a narrower term than 'crosslinguistic influence', although the two are often used interchangeably (Ringbom 2012).While in SLA this notion is used across different theoretical frameworks, views about linguistic transfer have undergone considerable change.Initially, it was assumed that the 'habits' of the L1 would be automatically carried over into the L2; now, scholars agree that L1 transfer works in complex ways and constitutes only one of the many factors and processes involved in L2 acquisition (Gass 1996).Research has shown that transfer-related differences apply mainly to early stages of learning: as learning progresses, all learners apply strategies and processes that are closer and closer to the TL (Ringbom 2012, 399).Traditionally, two types of transfer were distinguished: 'positive transfer' takes place when the influence of previous knowledge leads to flawless or rapid acquisition/use of new knowledge; 'negative transfer', on the other hand, occurs when such influence leads to errors or acquisitional issues.Recently, more types of transfer have been singled out: for example, transfer that predominantly happens from students' L1 (in our case Italian) or from other second/foreign languages the learner masters rather well (in our study, we take English as such a possible language), is called 'procedural' Negative procedural transfer may, in turn, either be intrusive or inhibitive.Intrusive transfer happens when learners inappropriately use structures or other items from their L1, while inhibitive transfer refers to L1 structures that inhibit learners to appropriately use new words and structures.This is well illustrated by Ringbom: TL words and structures without L1 parallels provide the learner with no concrete item transfer and are therefore often avoided as they are perceived as redundant.At one stage children learning L1 frequently produce forms like runned, goed for ran, went, thus avoiding what they apparently perceive as unnecessary redundancy for expressing past tense.Similarly, L2 learners initially also avoid what is perceived as redundant in the TL.(Ringbom 2012, 399; emphasis added) As we will see, both intrusive and inhibitive transfer are likely to be observed in this study.Crucial to the correct investigation of the role of linguistic transfer (both positive and negative) is a contrastive analysis of the differences between learners' L1 and L2, which is the focus of the next section.

Importance of L1-L2 Contrastive Analysis
A contrastive analysis of word order in students' L1 and L2(s) is important for several reasons.First, the role of cross-linguistic influence (Odlin 2012) is crucial in word order acquisition: studies have now attested that both the comprehension and production of a second language can be affected by cross-linguistic influence (Odlin 2012, 152).Moreover, studies show that, with respect to other L2 linguistic features, L2 word order is comparatively more influenced by L1 word order (James 1998).Pioneering studies on Chinese word order acquisition include the large-scale cross-sectional study conducted by Jiang (2009) on English L1 learners.Specifically, Jiang (2009) developed a taxonomy of Chinese L2 word order errors based on a contrastive analysis of Chinese and English, aimed at enabling explicit description and clear explanation of these errors.On the other hand, apart from a preliminary investigation conducted by Morbiato (2017), very few studies have so far been conducted on Italian L1 learners.Typological and grammatical differences existing between Italian and Chinese are often reflected in the errors students commit in conversation or translation tasks.Jiang ( 2009) and Morbiato ( 2017) present an overview of the salience of word order and its functions in the grammar of Chinese.Due to space constraints, I refer the reader to these studies, while in what follows a contrastive analysis is proposed with a special focus on double nominatives.As said earlier, in addition to Chinese and Italian, English is also considered as an example of students' already acquired second language: while much research concentrates on either L1-L2 influence, studies focussing on cross-linguistic influence involving a third language are a promising area of research (Odlin 2012).

Double Nominatives and Their Italian and English Counterparts. A Contrastive Form-Function Analysis
The double nominative construction is "typologically characteristic of many languages in the East Asian region" (Chappell 1996), as for example Korean and Japanese (Park 2010;Wunderlich 2014), while it is almost absent in the Indo-European languages of Europe.It has been referred to as "double-nominative" (Mullie 1933;Teng 1974a;Li 2004) or "double-subject construction" (Chappell 1996;Shen 1987;Modini 1981;Li, Thompson 1976, 1981;Tsao 1977) due to the feature of two NPs juxtaposed in sentence-initial position.In Chinese, it is referred to as 主谓谓语句 zhǔ-wèi wèiyǔ jù 'sentences with a subject-predicate predicate'.Investigation of this structure has been a continuing concern within Chinese linguistics: since the 1970s, a significant number of articles have been published on this topic, both by Chinese and e-ISSN 2385-3042  56, 2020, 379-410 ISSN 1125-3789

386
international scholars investigating Chinese grammar.In the China Academic Journals (CAJ) full text database, 386 articles can be found with the search term 主谓谓语句, 129 of which contain this term in their title.A paper such as the present one cannot do justice to all the literature on the subject: this section is devoted to a brief presentation of some of the main features such construction presents that are relevant to their acquisition with a particular focus on its structure (i.e., the form) and its semantic and discourse features (functions).

Structure
The double-nominative structure involves two nominals occurring adjacent to each other and followed by a predicative element, as in (1).Scholars tend to agree that this structure is related to the topicprominent nature of Chinese: Li and Thompson (1976, 480) consider it as a prototypical topic-comment sentence.According to such an analysis, the first NP is a topic, followed by a comment of the type NP-predicative element (PE).Since Chao's (1968) and Li andThompson's (1976, 1981) seminal works, the importance of the notion of topic in Chinese is well established in the literature: the fact that Chinese sentences are best described as topic-comment, rather than subject-predicate, structures is widely accepted across almost all theoretical persuasions. 5The topic is generally regarded as the sentence-initial element that 'sets a frame of validity for what comes afterwards' (Chafe 1976;Her 1991;Paul 2015;Morbiato 2018)  The comment is generally defined as what follows the topic and says something new about it.In this case, the comment is composed of the string NP2-PE.The predicative element located sentence-finally in the comment is often the focus of the message.Scholars have observed that, in many cases, the predicate is stative or intransitive (Chao 1968;Chappell 1996): the commonest subtype of an S-P predicate is one which describes the state or characteristic, less commonly an event, about the main subject.A personal subject may have an S-P predicate in which the subject represents a part of the body and the "small" predicate some physiological or psychological condition or property.(Chao 1968, 96) Recursivity Double nominatives can be recursive, thus constituting embedded topic-comment structures.One well-known such example is proposed by Her (1991, 6), who represents its structure as in figure 1; note again that English uses (and Italian would, too) either the possessive of-construction or the have-construction:   Another possibility is that of complex sentences where NP1 acts as a sentential or discourse topic that is later omitted to achieve textual cohesion, thus forming a topic chain (Tsao 1990;Li W. 2004Li W. , 2005)).In (6), the NP2 denotes a property of NP1, as observed by Chao above:

Semantic Features
Several scholars agree that the two NPs in the construction bear a semantic relationship between them, that is often that of wholepart, possessor-possessed (thing, property, characteristic), individual frame-characteristic, set-subset etc. 6 According to Modini (1981) and Chappell's corpus study (1996), such a relationship is that of inalienable possession in terms of the personal domain.Morbiato's corpus study (forthcoming) further reveals that inalienable possession is one of the instantiations of a more general linearization convention of the frame-part or containment schema, 7 that is observable for Japanese as well (Cook 1993).Morbiato proposes that Chinese displays a structural position, i.e. the preverbal position, and a relational structural pattern, i.e. the container-before-contained, or frame-before-6 See Chappell 1996 and Morbiato forthcoming for further discussion on the nature of such relationship.
7 Due to space constraints, the matter cannot be discussed in detail here (the reader is referred to Morbiato forthcoming); nonetheless, the nature of the containment relationship varies and must be interpreted metaphorically, in the sense of Lakoff and Johnson (2003).For example, NP1 in (6) 'that car' metaphorically contains its properties, such as its colour, its price, etc.

Anna Morbiato
Acquisition of Double-Nominative Constructions by Italian L1 Learners of Chinese part/participant, which inherently encode containment and frame setting.This is connected to the frame-setting properties of topics (Chafe 1976;Paul 2015;Morbiato forthcoming).In short, NP1 represents a semantic frame, a container within which something is predicated about NP2, that is a part/participant/property/contained thing.

Information-Structural Features
Chinese has a strong preference for 'initial topic' and 'end-focus'.The initial and final parts of the sentence are two very salient positions: as mentioned earlier, the former hosts the topic, i.e., the frame of interpretation, as well as the point of initiation, which contains given/identifiable/locatable information.These characteristics make it serve as a convenient introduction to the main point of a message (focus, new information) that resides in the comment.Given information carries very low Communicative Dynamism (CD), whereas new, focal information carries the highest CD (Firbas 1971).Thus, in the topic-comment structure, i.e., from the sentence-initial to the sentence-final position, the degree of CD tends to increase from low to high. 8It is common to process the information of a message in such a way as to achieve a linear presentation from low to high information value.This is referred to as the Principle of End-Focus (Quirk et al. 1985).Focus may be signalled in different ways, including prosodic prominence expressed by an accent or stress in spoken language, e.g., in Italian and English, while it is mostly signalled by the sentence-final position in Chinese (with exceptions as per the literature mentioned in fn. 6).By the Principle of End-Focus, the intonational nucleus in Chinese tends to fall on the last open-class lexical item of the last sentence element, in this case, the PE: The focus may also coincide with the whole comment (see Lambrecht 1994 andHole 2012 for a detailed discussion of different types 8 In fact, there are exceptions to this, including marked forms and marked focus constructions, focus markers (连, 也, 都, 才, 就 etc.), among others; however, this is generally not the case in the double nominative construction, hence I will not discuss such exceptions and refer the reader to Hole 2004Hole , 2012

Italian and English Influence. What Students Are Used to
As discussed earlier, Italian and English do not have a form as described in (8).How do these languages encode the three functions identified in (8.a-c)?English and Italian tend to encode containment and possession relationships lexically, rather than structurally, and namely: (i) with the verb to have [NP1 have NP2], as mentioned in the introduction; (ii) with possessive constructions (e.g., the of-construction in English and the preposition di in Italian, [NP2 of NP1].This is evident in the possible English and Italian translation of the Chinese sentences above.For example, possible Italian and English counterparts to sentences similar to (4) could be: 9. E. 'The flowers of that tree are very beautiful'.
This L1-L2 function-form difference is likely to have a negative impact on learners' interlanguage development: students may use the same patterns ([NP1 have NP2], [NP2 of NP1], or [NP1 adjectival predicate]) rather than the double-nominative construction (inhib-itive transfer).However, while these forms are indeed available in Chinese, and are grammatically correct/understandable by native speakers, the functions they express do not fully adhere to those of double nominatives.In what follows, each form is examined in detail.However, such pattern does not fully adhere to the end-focus principle.As discussed in section 3.1, focus in Chinese is normally searched in the last part of each sentence.In ( 11), the point is not that the subject has eyes, but rather, that his eyes are blue.The most salient, focal element is expressed by the adjectives (underlined), and not by NP2.However, the [NP1 have NP2] pattern involves that the sentence-final, and hence focal, element, is NP2.While this is not an issue in Italian and English, a sentence like (11.C) is less natural in Chinese.

NP1
The double nominative, on the other hand, allows the focal PE to occur in sentence-final, focal position.This is why a sentence like ( 7) is perceived as better, more natural, by native speakers (including those I surveyed).This is clearly stated by Loar: Normally verb elements cannot be focused in the SVO construction.However, when such focus is desirable, with an intransitive verb, like an adjective stative verb, End-Focus is easily achieved.Thus the S adjective-predicate sentence is widely used to highlight the [verbal] element that is communicatively prominent.(Loar 2011, 465) To illustrate this, Loar (2011, 465) proposes a number of sentence pairs (focal element is underlined): 12. E. 'There were truly lots of people on the plane'.Such a pattern also correctly allows the focal PE to occur sentencefinally.What is then the difference with (7)?And most importantly: how to explain this to students?The difference lies in what is to be analyzed as the topical element.In sentence ( 7), the topic is NP1 (tā); in sentence ( 13), NP1 modifies NP2 and forms with it a single constituent, that functions as the topic (tā de yǎnjing, 'his eyes').The choice lies in what the speaker wants the sentence (and the discourse) to be about.This can be better appreciated in longer strings of discourse: complex sentences such as ( 14) below may be useful for teaching purposes, as well.If we consider sentences (14.E-I), we notice that the topic of the discourse is he: the three clauses that follow all talk about that referent.A translation that employs the double nominative construction, as in (15.C), preserves the topicality of the referent, which is then omitted in its later occurrences, forming a topic chain.If, on the other hand, a 的 structure is used, as in (15.C'), the topic changes for each of the three clauses: tā de yǎnjing, 'his eyes', tā de tóufa, 'his hair', and tā, 'he', thus affecting the structure of the discourse and, in particular, discourse cohesion and topic continuity: However, sentence (16.C) has at least two problems: the first is that the adjective 稠密 chóumì should be referred to, e.g., people; hence a double nominative is in order, as in (16.C'); the second is its truth value: while it is true that China is populous (i.e., it has a big population), it is not equally true that it is 稠密 chóumì, i.e., densely populated (some areas, like its western parts, are definitely not).The correct, unmarked translation would then be a double-nominative construction of the type of (17).However, this involves a quite elaborate L1 form-function-L2 form transition, especially for beginnerlevel learners who may have not yet been exposed to this as a chunk/ formulaic expression.

China people very many
All in all, the double-nominative construction, while widespread in Chinese, is not available in languages like English and Italian, which instead use three other forms, including: and [NP1+adjectival predicate].These forms may therefore be preferred by students, in that they are more familiar.However, depending on the context, these three forms may present semantic, syntactic, and/or pragmatic/information structural differences as compared to the double-nominative construction.This may lead both to intrusive and to inhibitive transfer: students may use forms that are available in their L1, while usage of double nominative could be perceived as redundant and thus be 'inhibited'.This is what the study aimed at testing.The Study This section presents the details of the study, its hypotheses, participants, and data gathering method.The participants comprised three cohorts of Italian L1 learners of Chinese with different proficiency levels, for a total of 132 participants.Written production data, comprising translations of given sentences, was collected.Details are provided below.

Hypotheses
The study aims at verifying whether there is a correlation between L1-L2 differences with respect to form-function mappings and the percentage of the related word order errors in L2 production/comprehension due to Italian/English interference.Specifically, with respect to the present analysis, the test seeked to gather some empirical evidence to verify the following three hypotheses, based on the analysis in section 3: a. Double-nominative constructions.Italian L1 students are not familiar with this construction, in that neither Italian nor other Indo-European languages that are generally studied in Italian high schools (English, French, Spanish, German) do allow such patterns.Hence, Italian L1 learners might find it difficult to acquire it and, consequently, produce it.b.Encoding of frame-part/participant/property, containment, and possession relationships.Given the L1-L2 differences in expressing such relationships, learners -and especially beginner level ones -may tend to produce structures that are similar to those of their L1, and namely [NP1 have NP2], [NP2 of NP1], or [NP1 adjectival predicate].c.Topic-first and End-focus.while Italian and other European languages tend to follow the universal tendency of given-tonew, topical-to-focal information progression, they tend to be less rigid than Chinese.Thus, Italian L1 learners, just like English L1 learners (Jiang 2009), might not feel the need to adhere to Chinese topic-first and end-focus patterns.As topic is, again, not as relevant in the grammar of Italian or English, topic-first related word order error rates may tend to be high also among advanced-level students.
All hypotheses (i-iii) involve negative transfer/inference, and thus a relatively high percentage of word order errors is expected in related translation tasks, especially in lower proficiency levels and in more complex sentence types -e.g., with embedded double nominatives of the type of (5).

Test Design
For the study, a cross-sectional design was chosen over a longitudinal one.While a longitudinal design seeks to obtain data on the development of interlanguage and L2 over time, and for a small number of learners (generally one), a cross-sectional design gathers data at one point in time, and from learners at different developmental stages. 10A cross-sectional design was chosen because it allows gathering a statistically relevant amount of data from a large number of Chinese L2 learners at different proficiency levels, who are more likely to produce a greater and more varied number of errors.The test consisted of an Italian-to-Chinese translation task.It was designed to test only structures and word order patterns; hence all the necessary vocabulary was provided in brackets.The test was anonymous to prevent students from feeling the pressure of making mistakes.Data considered for this research comprise three sentences which native speakers would translate using simple double-nominative constructions, i.e. ( 18), ( 19) and ( 20), and one sentence which native speakers would translate using a double-nominative construction embedded in a relative clause, i.e. ( 21).These sentences were submitted to a control group of 5 Chinese native speakers, all with university-level education, 3 of which are Chinese language teachers, all born in China.Sentences are reported below, along with translations provided by native speakers:

Data Collection and Participants
The test was submitted to three cohorts of Italian L1 learners of Chinese with different proficiency levels -and a focus on lower levels for the reasons discussed above, for a total of 132 participants.Group A is the group with the lowest proficiency level: it comprises 30 students enrolled in the 2nd year of the BA degree in Languages and Cultures for Tourism and International Commerce, University of Verona.
In their first year, students received 3 hours (grammar) + 3 hours (conversation) of lessons per week in both semesters; 5 of them had studied Chinese before.Group B has a slightly higher level of proficiency: it comprises 54 students enrolled in the 2nd year of the BA degree in Foreign Languages and Literatures and Languages, Markets and Cultures of Asia and Mediterranean Africa, University of Bologna.In their first year, students received 3 hours (grammar) + 4,5 hours (conversation) of lessons per week in both semesters; 4 of them had studied Chinese before, and 8 of them had spent a few weeks in China.Students in Group C are the most advanced: the group comprises 49 students enrolled in the 1st year of the MA degree course in Editorial Interpreting and Translation at the University of Venice, Italy.48 of them hold a bachelor's degree in Chinese or related fields from 12 different Universities, hence they constitute an interesting and diverse sample, representative of BA-level Italian L1 graduates majored in Chinese.One student is a Chinese native speaker who got his MA from the Chengdu Institute of Sichuan International Studies University in China, and is one of the members of the control group, as discussed in section 4.2.The test was submitted at the beginning of the semester, to avoid interference with contents and knowledge shared in the upcoming teaching unit.

5
Results.Quantitative and Qualitative Analysis

Quantitative Analysis
The following tables show rates and percentages of error types grouped according to the three students' cohorts.The tables below, on the other hand, gather data from different learners' cohorts for the same sentence.Again, figures indicate how many times each sentence was translated using the indicated form: IV. First sentence: usage rate for each form The Chi-Square result shows that there is a significant difference between the answers given within the three groups (χ² (8) =153.77,p = 3.19E-27 < 0.05).Also, the Chi-Square calculated on doublenominatives vs. other forms show significant difference (χ²(2) =9.09, p = 0.01 < 0.05).This suggests that there is a significant difference of performance between the three groups.From the above data, an overall reorganization of the balance of means employed by students can be observed: group A tends to use the 有 structure (i.e., the NP have NP2 pattern) the most, while the 的 structure (of specification pattern) is increasingly preferred in all sentences as the proficiency level rises (e.g., 10% vs. 75% in the third sentence and 13.33% vs. 85.2% in sentence 4).Nonetheless, the double-nominative structure is used rarely at all proficiency levels, although higher rates are observable in Group C.

Other Errors
Finally, the analysis has highlighted other types of errors.Sentence (32) seems to treat hǎo píqi as an adjective -just as in (27-29) above -that is placed in a 是……的 shì…de construction (maybe to provide emphasis or because it resembles an individual-level predicate that requires the copula shì, 'be', just like adjectives such as 方 fāng, 'squared'):  25).The [NP1 的 NP2] pattern is also very widespread and used across all sentence types and proficiency levels.Errors of this type do not compromise the intelligibility of the sentence, nor do they 'feel wrong' in isolation.Maybe this is the reason why the usage rate of this structure increases across all sentence types from lower to higher proficiency levels.The [NP adjectival predicate] pattern is used mostly for sentence 4, which confirms the role of L1 influence and word-for-word translation: a significant number of students left the adjective in its Italian form 'popoloso', others used semantically wrong adjectives, e.g., chóumì, 'dense'.In (59) a synonym of dense, i.e., 挤 jǐ, 'crowded', is used.However, as said earlier, the truth value of the sentence is compromised, as China is not the most densely populated/crowded country in the world.Overall, a both intrusive and inhibitive transfer connected to L1 forms can be hypothesized to cause the low rates of production of the target form, i.e., double nominatives.

Conclusions
While the double-subject construction may appear as a simple and straightforward pattern, the present study shows that it is in fact a rather difficult construction to acquire for Italian L1 learners.Rather, students tend to use patterns they are used to in their L1 (or other L2s, such as English).Similar results were obtained in Jiang ( 2009): English L1 students also find it difficult to naturally produce this pattern, along with other topic-comment related constructions.Often, L2 word order errors are the result of learners relying on carrying out word-for-word translations of native language surface structures when producing written or spoken utterances in their target L2 performance.This is what seems to be happening in most cases.L1 word order features provide one of the most important sources for adult learners' L2 word order errors.Hence, topic-comment related constructions such as double nominatives need to be introduced to students, not only for theoretical accuracy, but also to clarify some structural and functional peculiarities of Chinese sentences as compared to, e.g., English and Italian.In particular, the functions of double nominatives (i.e., allowing focal elements to occur sentence-finally and topical ones to occur as the first constituent in the sentence) should be introduced when explaining this pattern.Examples of the pattern should be explained and compared to forms (translations) in students' mother tongues, explaining similarities and differences, as well as why a form is preferred over the other.
flowers of this tree have very nice colours'.

Figure 1
Figure 1 Representation of multiple double nominative constructions as embedded frames (adapted from Her 1991)

Anna Morbiato Acquisition of Double-Nominative Constructions by Italian L1 Learners of Chinese Anna Morbiato Acquisition of Double-Nominative Constructions by Italian L1 Learners of Chinese
Chafe 1976;Paul 2015 andMorbiato 2018t seeChafe 1976;Paul 2015 andMorbiato 2018for a critique of this generalization).The topic may, but need not, be argument of the predicative element.Accordingly, N1 (the topic) is not a verbal argument: the sole argument of the PE jìnshì 'short-sighted' is NP2, i.e., yǎnjīng 'eye': 5 An overview of the historical development of this notion for Chinese is beyond the scope of the present article; see, among others, Li 2005, Shyu 2016 and LaPolla 2016.Annali di Ca' Foscari.Serie orientale e-ISSN 2385-3042 56, 2020, 379-410 ISSN 1125-3789 That car, the price is too high, I don't want to buy it'.
and Xu 2015 for further details.i.e., sentential, predicate and narrow focus).To sum up, the form-function association connected to double nominatives involves: (c) End-focus: Saliency of the sentence-final element(s).

有 (have, exist) NP2.
If we consider a sentence like (7), it is in fact possible to express it by preserving the English/Italian [NP1 have NP2] structure, namely with the 有 (have, exist) verb: Specifically, figures indicate how many times each sentence was translated using the indicated form: I. Group A: students' translations

Anna Morbiato Acquisition of Double-Nominative Constructions by Italian L1 Learners of Chinese Anna Morbiato Acquisition of Double-Nominative Constructions by Italian L1 Learners of Chinese
Venice has a small population') in group C may be related to the successful acquisition of the formulaic expression 中国人口很 多, which students have very likely already been exposed to at that level of proficiency.This hypothesis is preferred to that of a successful acquisition of the double-nominative construction in that double nominative percentages are low for the same students in simpler sentences like sentence 1 and 2. The [NP1 have NP2] pattern is used across all sentence types and proficiency levels, hence it is one of the errors students tend to stick with throughout their learning process, although it decreases over time through interlanguage development.It is worth noticing that L1 transfer is present not only with respect to the [NP1 have NP2] pattern, but also in the word-for-word translation of the Italian indefinite article 'un' (English 'a'), that is expressed by the numeral classifier string 一个 in a number of sentences, e.g. (

Anna Morbiato Acquisition of Double-Nominative Constructions by Italian L1 Learners of Chinese Anna Morbiato Acquisition of Double-Nominative Constructions by Italian L1 Learners of Chinese
This section includes examples of the most relevant errors committed by students.