Language interaction effects in bimodal bilingualism: Argument omission in the languages of hearing ASL-English bilinguals

Revised Draft July 2015

1

Language interaction effects in bimodal bilingualism: Argument omission in the languages of hearing ASL-English bilinguals

Elena Koulidobrova∗

Central Connecticut State University 1615 Stanley St, FD 208-12

New Britain, CT 06050 [email protected]

Abstract: The focus of the paper is a phenomenon well documented in both monolingual and bilingual English acquisition: subject and object omission. Previous studies have shown that bilinguals acquiring a null and a non-null argument language simultaneously tend to exhibit unidirectional cross-language interaction effects—the null argument language remains unaffected but oversuppliance of overt elements in the null argument language is observed. Here subject and object omission in both ASL (null argument) and English (non-null argument) of young ASL-English bilinguals is examined. Results demonstrate that in spontaneous English production, ASL-English bilinguals omit subjects and objects to a higher rate, for longer, and in unexpected environments when compared with English monolinguals and bilinguals; no effect on ASL is observed. Findings also show that the children differentiate between their two languages—argument omission rates during ASL- vs. English target sessions differ. Implications for the general theory of bilingual effects are offered. Keywords: null subjects, argument omission, bimodal, sign-speech bilingualism

1. Introduction

Recently, a new dimension has entered the discussion of bilingual experience: unimodal vs. bimodal bilingualism. The former refers to an ability to use more than one language in the same modality—i.e. spoken-spoken or signed-signed, as in the classic understanding of bilingualism (e.g. Spanish-English, Japanese-French, American Sign Language (ASL)-French Sign Language (LSF), i.a.). The latter describes the knowledge of two languages in two different modalities—spoken and signed (e.g. English-ASL, Japanese-LSF, i.a.). On the assumption that sign languages are, in fact, natural languages (Sandler & Lillo-Martin, 2006), examination of

∗ I gratefully acknowledge the contribution of bimodal bilingual participants and their families. This research is supported in part by Award Number R01DC009263 from the National Institutes of Health (National Institute on Deafness and Other Communication Disorders). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIDCD or the NIH. Support is also provided by awards from the Gallaudet Research Institute and from CNPq (Brazilian National Council of Technological and Scientific Development) Grant #200031/2009-0 and #470111/2007-0. I am thankful to all of the collaborators and research assistants for their effort and feedback at various stages of the project.


2

language acquisition by bimodal bilinguals has potential to offer a particular contribution to bilingualism research: if, for instance, bimodal bilinguals develop in the manner documented for unimodal bilinguals, then bilingualism effects (whatever they might be) are modality insensitive. However, if bimodal bilinguals behave differently, then a contribution of (cross-)modality must be accounted for. This paper focuses on one instance of bilingualism effects in one type of bimodal bilingual population—young hearing children of Deaf adults (a.k.a. kodas1).

Linguistic patterns of this population have been examined by various researchers over the past three decades (Todd, 1971; Schiff & Ventry, 1976, Sachs et al., 1981; Murphy & Slorach, 1983, Schiff-Myers, 1988; Johnson et al., 1992; Seal & Hammet, 1995; Messing, 1999; Marshall et al., 2005; Pizer, 2008, Chamberlain & Mayberry, 2008; Kovelman et al., 2008; Kovelman et al., 2009; Bishop, 2009; Jarque, 2010). Studies concur that the findings mirror what is known about unimodal bilinguals: for instance, children exhibit parallel lexical growth in sign and spoken languages, produce “translation equivalents,” and are sensitive to the language of the interlocutor (cf. Petitto et al., 2001; Petito & Holowka, 2002; Holowka et al., 2002; Petitto & Kovelman, 2003; Brackenbury et al., 2006); their linguistic patterns are affected by the nature of the input (Capirici et al., 2002; van den Bogaerde & Baker, 2005). Researchers have also noticed that both spoken and sign languages of this population often show what may be described as instances of ‘incorporation of grammatical properties’ from the other language. These effects have been recorded for at least five language pairs: ASL-English (cf. Todd, 2008; Lillo-Martin et al., 2010), Brazilian Sign Language (Libras)-Brazilian Portuguese (Lillo-Martin et al., 2010; et seq.), Italian Sign Language (LIS)-Italian (Donati & Branchini, 2009), the Sign Language of the Netherlands (NGT)-Dutch (van den Bogaerde & Baker, 2005) and British Sign Language (BSL)-English (Morgan, 2000). The focus of this study is a phenomenon well documented in both monolingual and bilingual English acquisition—argument omission a null- (ASL) and a non-null (English) argument languages. The findings of a longitudinal study demonstrate ASL-English bilinguals behave differently from their monolingual and unimodal bilingual comparisons.

2. Previous research: Argument omission in English 2.1 Monolinguals English falls into the class of languages forcing arguments to be overt, although in some cases, argument omission appears possible. These null arguments (NAs), however, are allowed only under restricted syntactic, pragmatic, and semantic conditions: subjects may be omitted at the left edge—i.e. everything to the left of them must be phonologically empty (cf. Sigurðsson & Maling, 2009; Sigurðsson 2011)—and, typically, when in reference to the 1st-person singular (1SG). For example, the auxiliary ‘do’ is generally compatible with either 1SG or any 1st-/ 2nd-

/3rd-person plural (1-/2-/3PLU), and ‘should’ with any person or number. Yet, in (1), only one

1 Following Lillo-Martin et al. (2009), i.a., I refer to young hearing children of Deaf adults as ‘kodas’, vs. adult hearing children of the Deaf—codas.


3

reading is available—1SG, and the sentence is grammatical only if the left-most element is omitted. (1) a. ___ Don’t think we can make it tonight.

= I don’t think… ≠{We / you / they} don’t think…

b. __ Should really go to the gym tomorrow. (adptd. Weir, 2009) = I should… ≠{We / you / they / she / he} should…

c. *Tomorrow __ should really go to the gym. Objects of transitive verbs, such as buy, read, and hear, can also be omitted; however, as is briefly demonstrated in (2), these arguments (a.k.a. ‘implicit’) are clearly the property of the verbs themselves, not some grammatical phenomenon in the language (cf. Bhatt & Pancheva, 2006; AnderBois, 2011). For example, the omitted argument of buy (2b) has an indefinite reading only, and the argument of hear (2c) disallows the indefinite interpretation.

(2) a. Let’s go out. Mary is buying ({ok drinks / *the drinks}) today. b. She is reading ({ok something / *the news}). c. Have you heard ({ * anything /ok the news})?

Overall then, English is not a language that allows argument drop. Yet, young English-speaking children omit subjects more than it seems they should. The study of argument omission in monolingual English learners boasts a long history and has served as a vehicle for a variety of proposals for the general view of early grammar. For instance, argument omission/suppliance rates have served as evidence for discoveries about (i) processing difficulties in linguistic performance (e.g. Bloom, 1990; Boster, 1997), (ii) metrical effects in child language (e.g. Gerken, 1991), (iii) parameter triggering mechanisms (cf. Borer & Wexler, 1987; Roberts & Holmberg, 2010, i.a.), (iv) the amount of structure initially available to the child and how much of it ‘comes on-line maturationally’ (Borer & Wexler, 1992; Borer & Rohrbacher, 2002; Hyams, 1992, i.a.); (v) the role of information structure in children’s language (Allen, 2000, Guerriero et al., 2001, Hughes & Allen, 2008, i.a.), and many others. The studies reported here will not shed light on any of the aforementioned issues. What will become relevant is that among the findings in the monolingual literature are a) root position effects (Roeper & Weissenborn, 1990; Valian, 1991, i.a.); b) dissociation of argument omission rates in the English-type from the Italian-type languages (Lillo-Martin, 1991, Valian, 1991); and c) correlation effects between argument omission and the systematic correct suppliance of tense/agreement morphology (cf. Guilfoyle, 1984; Sano & Hyams, 1994; Hoekstra & Hyams, 1998). Although the results (as well as methodologies) of studies cited above vary, it is by now well documented that in spontaneous production, monolingual English children exit the stage in which they incorrectly omit arguments (primarily subjects) by the age of 3 and MLUw>3, more or less concurrently with


4

having demonstrated the knowledge of morpho-syntax, as evidenced by knowledge of C-domain related processes (see Hyams, 2011 for a detailed historical overview of the inquiry). While they often omit (~50%) required subjects and objects (~5%) when younger, at this stage of development, the rate of subject omission tends to hover well below 10%; the rate of object omission is typically recorded to be below 1% and non-existent after the age of 3. Moreover, at any age, although subjects may be missing at the left edge (as in (1)) and in non-finite clauses, they never are in finite embedded clauses; null objects are always 3rd person singular. This sharply contrasts with parallel data from languages like Italian (see Serratrice & Sorace, 2003 for an overview) and Chinese (Wang et al., 1992), where children’s rates of argument omission mirrors the adult rates—typically above 50%. Since this paper examines argument omission in spontaneous production of children for whom English is a first language, we might expect them to pattern in the manner described above.

2.2 Bilinguals Precisely because much monolingual data are now available, it is tempting to draw a comparison between monolinguals and bilinguals in order to bring into focus processes characteristic of and unique to bilingualism. Serratrice, Sorace, & Paoli (2004) examine the rates of argument omission in the two languages of a balanced (45% of waking hours per language) Italian-English bilingual named Carlo (age 1;10-4;06) and illustrate that he converges on the target English grammar in terms of subject and object suppliance even before monolingual controls do. They isolate four stages of linguistic development based on the MLU: Stage 1 (MLUw 1.5-2.0), Stage 2 (MLUw 2.0-3.0) Stage 3 (MLUw 3.0-4.0), and Stage 4 (MLUw > 4.0) and show that at the stage of development typically associated with the knowledge of the relevant aspects of English morpho-syntax (MLUw>3, age 3), when monolingual English speakers exhibit mastery of the language’s requirement for the (i) overt subject of the clause, and (ii) overt object/complement if one is required in the adult grammar, Carlo behaves similarly: although Italian subjects are allowed/preferred to remain null (unless in shifted topic scenarios, see Frascarelli, 2007), Carlo begins supplying subjects in English correctly very early, in line with, or even earlier than, monolingual English children.

With respect to null objects, English and Italian differ to some degree as well: neither language allows a productive object omission (barring English ‘implicit arguments’ as in (2)). However, Italian allows null objects in at least one form: proarb (Rizzi, 1986).

(3) La buona música riconcilia __ con se stessi. The good music reconciles with self ‘Good music reconciles __ with oneself.’ (Rizzi, 1986) Therefore, it is at least in principle possible for both null subjects (NSs) and null objects (NOs) —or, perhaps, the property of grammar resulting in each – to transfer into Carlo’s English from


5

his Italian. However, the rate of NOs in Carlo’s English does not differ from monolinguals’. Moreover, as Serratrice et al. report, the rare cases of null arguments in Carlo’s English are consistent with the English grammar; i.e. they are not qualitatively different either. In other words, in his English, Carlo performs in line with monolinguals. The data for Carlo as well as the monolingual (Brown, 1973; Suppes, 1974; and Sachs, 1983, respectively) comparisons from Serratrice et al. are in Tables 1-2 (subject and objects omissions, respectively). Table 1: Rates of subject omission in Carlo’s English as compared to monolinguals (Serratrice et al. 2004) Stage CarloIt-Eng Adam Naomi Nina Sarah ENGMean

1 4/32 (.12) --- 17/78 (.22) 128/248 (.52)

36/80 (.45) 181/406 (.45)

2 51/700 (.07)

163/411 (.39)

32/109 (.29) 101/555 (.18)

30/174 (.17)

323/1249 (.26)

3 9/304 (.03) 22/520 (.04)

13/162 (.08) 78/755 (.10)

19/282 (.07)

132/1719 (.08)

4 8/461 (.02) 41/1004 (.04)

18/321 (.06) ---- ---- 59/1384 (.04)

Table 2: Rates of object omission in Carlo’s English as compared to monolinguals (Serratrice et al. 2004)

Stage CarloIt-Eng Adam Naomi Nina Sarah ENGMean

1 --- --- 0/22 (0) 0/42 (0) 4/41 (.10) 4/85 (.04) 2 2/186 (.01) 8/194 (.04) 3/47 (.06) 10/226

(.04) 7/96 (.07) 28/563

(.05) 3 3/89 (.03) 6/228 (.03) 3/85 (.03) 4/270 (.01) 0/107 (0) 13/690

(.02) 4 2/176 (.01) 1/380

(.003) 1/66(.01) --- --- 2/446

(.004)

The findings in Serratrice et al. have been supported by studies in various language combinations, at least one of which is a null argument (NA) language, irrespective of the type of NA (cf. Roberts & Holmberg, 2010) the language is argued to have (cf. Juan-Garau & Pérez-Vidál, 2000 [Spanish-English]; Serratrice et al., 2004 [Italian-English]; Pinto, 2006 [Italian-Dutch]; Hacohen & Schaeffer, 2007 [French-English, Hebrew-English]; Schmitz et al., 2012 [Italian-German, French-German]; Mishina-Mori, 2007 [Japanese-English]; Blaise et al., 2010 [Japanese-French]; Huang, 1999 [Chinese-English]; Haznedar, 20010 [Turkish-English]; Zwanzier et al., 2004 [Inuktitut-English]). Although varying in methodology, the studies above demonstrate no visible effect on the bilingual child’s non-NA language with respect to subject


6

omission. On the other hand, many, Serratrice et al. included, record oversuppliance of overt arguments in the bilingual’s NA language. The reasons for this remain under debate (Serratrice et al., 2004; Sorace, 2011; Tsimpli, 2011; Liceras et al., 2012; Pinto, 2013; i.a.), but one thing is clear: unimodal bilingual studies in general have illustrated that children acquiring a NA and a non-NA language simultaneously show no delay in ‘exiting the NA stage’ in their non-NA language—i.e. they do not show cross-linguistic ‘transfer’ in the domain of argument omission. At least in their non-NA language, bilinguals are reported to perform in line with monolinguals. One way of interpreting such data is this: to the degree that cross-linguistic interaction occurs, it does so unidirectionally—from NA- to the non-NA language. 3. Study: ASL-English bilinguals 3.1. Predictions The language pair under discussion here is ASL-English. Unlike English, and like many Romance and East Asian languages, ASL allows and often prefers, as indicated by the asterisk on overt pronoun IX in (4), nominal arguments to remain silent (Bahan et al., 2000; Lillo-Martin, 1991; Koulidobrova, 2012): (4) a. A: Have you seen my candy?

B: YES, { __ /1IX} EAT-UP { __ /*a-IX} ‘Yes, (I) ate (it).’ (adptd. Lillo-Martin, 1991) b. ARTHUR WORRY { __ /*a-IX} WILL FALL ‘Arthur is worried that (he) will fall.’

Native ASL signing children acquire this property early. Lillo-Martin (1991), Quadros et al. (2001) and Quadros & Lillo-Martin (2007) demonstrate that between the ages of 1;08-2;10, Deaf signers omit over 75% of subjects and over 10-70% of objects (depending on the verb-type) in their ASL. At ages 3:06-5:09, rates of argument omission in monolingual ASL signers drop significantly (Lillo-Martin, 1991). Unfortunately, Lillo-Martin does not offer the exact numbers for comparison, and no other study of argument omission in native Deaf users of ASL of comparable ages currently exists (to my knowledge). Adult studies record 35-45% of argument suppliance (Wulf et al. 2002). Thus, comparison of rates of argument omission in ASL of kodas 3-6 years of age is not possible: if, for instance, they omit subjects more, or less, than 55-65% observed in the adult language corpus, their argument suppliance rates may be still be consistent with the monolingual child data at the relevant stages of linguistic development. What can be quantified, however, is the grammaticality of their argument suppliance, as compared with adult language, as well as the difference between the rates of argument omission in koda ASL vs. their English. Judging from the previous studies on unimodal bilingualism, effects are expected in only one direction—in the NA, but not the non-NA, language. We might then formulate the following hypotheses for the ASL-English language pair: whatever the account… (5) a. The rate and contexts (i.e. syntactic environment) of argument omission in the non-NA

language of ASL-English bilinguals (English) will be comparable with that of


7

monolinguals and unimodal English bilinguals. b. The rate of argument omission in the NA-language of ASL-English bilinguals (ASL) will be lower than what is observed in monolingual ASL production; overt arguments will be supplied in the environments dispreferred by the ASL adult grammar.2

In other words, ASL-English bilinguals are expected to perform like Carlo in Serratrice et al. (2004), taken here to represent other studies on cross-linguistic interaction in the domain of argument omission in in similar language pairs. 3.2. Subjects and Methodology In this study, I examine the English of two young kodas: TOM and LEX from the BIBIBI project (Lillo-Martin et al., 2009; Chen Pichler et al., 2010). Both children, each with at least one Deaf signing parent, are being raised bilingually. Input in English is provided by a number of hearing family members as well as the English-based daycare and playgroups which both TOM and LEX began to attend before the age of 5 months.

Children were filmed biweekly either at home, daycare, or a play session at the Gallaudet Research Institute, interacting with either Deaf parents or familiar Deaf researchers during ASL sessions and with native English speaking family members or familiar researchers, all with knowledge of ASL, during the English sessions. Sessions range between 35-50 min. Data were transcribed, coded and analyzed using ELAN (http://www.lat-mpi.eu/tools/elan/), following the conventions established by Chen Pichler et al. (2010), many of which are parallel to those used in CHILDES (MacWhinney, 2004).

All non-linguistic utterances (i.e. vocalizations, xxx, yyy), repetitions and direct imitations were excluded from the analysis. All utterances subject to analysis were coded according to their compatibility with the adult language in a given context: CAL (consistent with adult grammar) vs. NAL (non-adult grammar). A special tier was created in ELAN for coding and analysis of 2 Note that (5) does not appear to be based on any particular theoretical framework regarding the nature of the potential specific language interaction effects in the languages of ASL-English bilinguals; rather, it reflects the empirical observations from previous studies of comparable language combinations. This is due—in part—to the fact that the influential model of ‘transfer’ between the languages of a bilingual (Hulk & Müller, 2002, i.a.) does not easily accommodate the findings. (i) a. Cross-linguistic influence occurs at the interface between two modules of grammar, and more particularly at the interface between pragmatics and syntax in the so-called C-domain[…]. b. Syntactic cross-linguistic influence occurs only if language A has a syntactic construction which may seem to allow more than one syntactic analysis and, at the same time, language B contains evidence for one of these two possible analyses […] (Hulk & Müller, 2000:228-229) Concretely, the model in (i) predicts that English-Spanish/-Italian/-Turkish/-Chinese/-Inuktitut bilinguals would omit arguments in their English, simply because all of the aforementioned languages exhibit argument omission properties governed by both syntax and pragmatics. Therefore, both conditions of (i) ((a) and (b)) are satisfied and cross-linguistic influence should be observable. For the moment, I will set (i) aside but will return to it in the discussion section of the paper.

Serratrice et al. (2004) defend a different hypothesis of potential language interaction effects based on discourse appropriateness (Allen, 2001). Examination of the data presented here in terms of the relevant informativeness features is reported elsewhere (Author, in submission). The focus of this study is the rate and syntactic environments of argument omission.


8

utterances containing verbs; only such utterances were subjected to the analysis. These were further coded as “_ / _ ”, with the first slot reserved for a number indicating whether the verb required an overt subject (e.g. “1/_” if finite and “0/_” if imperative) and the second slot reserved for a number indicating obligatoriness of an overt complement (e.g. “_/1” if unambiguously transitive, like ‘kiss,’ and “_ /0” if intransitive or allowing the argument to remain ‘implicit’). Recall that English allows null arguments in a restricted set of cases (such as (1)-(2)); consequently, if an argument was omitted in such an environment, the resulting utterance was considered target-like (CAL). If, however, in a given context an overt subject (e.g. with finite verbs) or object (e.g. with unambiguously transitive verbs requiring an overt complement in adult language) was required, the utterance was coded as non-target (NAL). Independent tiers were created for null subjects and objects: “1” if the relevant argument was omitted.

The omitted argument was then coded for the implied person/number as appropriate from the context (1SG, 1PLU, 2, 3SG, 3PLU). Finally, a subset of data was analyzed for reference shift from previous linguistic context (e.g. the topic to be maintained is 3SG but the null argument is 1SG). The coding decisions above were initially checked by 1-3 native speakers; additionally, 10% of total utterances were subjected to a reliability check by a independent trained coder, with 99% agreement attained. An example of the ELAN coding window is shown in Figure 1.

Figure 1. ELAN window.

Following the methodology offered in Serratrice et al. (2004), four stages of linguistic development were isolated: Stage 1 (MLUw 1.5-2.0), Stage 2 (MLUw 2.0-3.0) Stage 3 (MLUw 3.0-4.0), and Stage 4 (MLUw > 4.0). MLUw was calculated based on the guidelines from Brown


9

(1973). ASL MLU is not reported for various reasons, one of which is that it reportedly does not increase with language development, making it difficult to use it as a predictor (Berke & Lillo-Martin, 2012).

The largest gap between sessions is 6 months for LEX (between stages 2 and 3) and 5 months for TOM (at both edges of stage 3). Descriptive data for each child respectively are recorded in Tables 3-4. Table 3. TOM Stage Nu. of

transcripts Age MLUw

English Total nu. of English utterances

Total nu. of ASL utterances

1 3 1;11.21-2;03.13

1.2-1.96 239 -------

2 5 2;04.15-3;03.00

2.14-2.55 1426 202

3 8 3;05.08-4;05.01

3.03-3.45 2677 187

4 3 4;06.03-4;11.09

4.09-4.3 1459 143

Total 19 1;11.21-4;11.09

1.2-4.3 5801 532

Table 4. LEX Stage Nu. of

transcripts Age MLUw

English Total nu. of English utterances

Total nu. of ASL utterances

1 ------ ------ ------ ------ -------- 2 1 3;03.12 2.91 336 --------- 3 4 3;08.27-

4;03.11 3.1-3.75 1936 74

4 4 4;05.13- 4.5 2473 563 5;04.06.20 Total 9 3;03.12-

5;04.06 2.91-5 4745 637

At each stage of linguistic development the rates of NSs NOs in TOM and LEX’s English

(individually and as a mean) were compared to that of the Italian-English bilingual Carlo and English monolinguals Adam, Sarah, Naomi and Nina (Brown, 1973; Suppes, 1974; and Sachs, 1983, respectively).

3.3 Part 1: English-target 3.3.1 Data


10

In this part of the study, English-target sessions only are examined. The hypothesis tested in this part of the study is (5a), repeated here as (6). (6) The rate and contexts (i.e. syntactic environment) of argument omission in the non-NA

language of ASL-English bilinguals (English) will be comparable with that of monolinguals and unimodal English bilinguals. = (5a)

Concretely, this means that in English, TOM and LEX may be expected to omit arguments in the contexts monolingual English learner do, but by the age of 3 and MLUw>3, they will have ‘exited the NA-stage’—i.e. at Stages 3-4, argument omission in English is unexpected. In addition, if TOM and LEX significantly differ from Carlo, (i) this will serve as evidence that ‘cross-linguistic influence’ in the domain of argument omission/suppliance across language combinations is not unidirectional, as might seem to be the case from the unimodal bilingual data, and (ii) an account will be required which would explain this difference—something along the lines of the unique characteristics of bimodal, vs. unimodal, bilinguals.

Tables 5-6 show that as expected, both TOM and LEX produce NSs and NOs in their English; the raw numbers and proportions of NS and NO (93% of which were nominal, 6% may be interpreted as VPE) per child and per stage are provided.

Table 5. TOM, English

Stage Nu. of utterances with verbs

Nu. of NS (proportion)

Nu. of utterances requiring verbal complement

Nu. of NO (proportion)

Total nu. of utterances

1 42 11 (.26) 28 2 (.07) 239 2 3 4

509 1102 504

92 (.18) 131 (.12) 46 (.09)

341 892 430

24 (.07) 28 (.03) 23 (.05)

1426 2222 1045

Table 6. LEX, English






1 2 3 4

----- 177 1058 533

----- 49 (.28) 123 (.12) 33 (.06)

----- 168 984 470

----- 17 (.11) 71 (.07) 8 (.02)

----- 336 1134 1179

The data were collapsed across the two children; the mean (henceforth KODAMean) is provided for each stage in Table 7. Table 7. KODAMean, English


11






1 42 11 (.26) 28 2 (.07) 239

2 3 4

626 2160 1037

141 (.23) 254 (.12) 79 (.08)

502 1876 900

41 (.08) 99 (.05) 32 (.04)

1762 4155 2224

The methodology of Serratrice et al. (2004) was followed: (a) at each stage of linguistic development, the proportions of NSs in TOM and LEX’s English were compared to that of the Italian-English bilingual Carlo and English monolinguals Adam, Sarah, Naomi and Nina (Brown, 1973; Suppes, 1974; and Sachs, 1983) individually and as means; (b) difference in rates was calculated. Results are reported in percentages for ease of exposition and are graphically represented in Figures 2-3. Examples of NSs at Stages 3-4 are provided in (7)-(8) and NOs in (9)-(10).

Figure 2. NSEng, English sessions Figure 3. NOEng, English sessions

(7) TOM:

a. Inv: Put it in like that and it will be straight. CHI: it’s not straight like that Inv: Trust me CHI: ___ build house b. Inv: It’s a window. You are right. CHI: This is gonna be a cool.

0

10

20

30

40

50

Stage 1 Stage 2 Stage 3 Stage 4

% o

f NS

in u

ttera

nces

con

tain

ing

verb

s (an

d re

quiri

ng su

bjec

ts)

Subject omission (means): English

KODA ENG Carlo

0

10

20

30

40

50

Stage 1 Stage 2 Stage 3 Stage 4

% o

f NO

in u

ttera

nces

con

tain

ing

verb

s (a

nd re

quiri

ng o

vert

obje

cts/

com

plem

ents

)

Object omission (means): English

KODA ENG Carlo


12

Inv: It is going to be cool. Yeah. CHI: Can ___ give me this? c. CHI: He is fast (talking about a car) Inv: &=imit:car CHI: ___ have to build this CHI: It says ____ have to build a king of the king horsies d. CHI: How did ___ tape it? (8) LEX:

a. CHI: Hmmm, it goes right over there. CHI: I think the sheep one might work CHI: ___ think this one go with this. b. CHI: You got to put it down the very last one Inv: Awesome CHI: We made it into puzzle &=coughs CHI: ___ have a very bad[?] cold Inv: You have a cold? c. CHI: And now I’ll stay my mom. Inv: That’s right. CHI: It’s gonna pretty soon. MOM DAD SOON Inv: Pretty soon. CHI: Pretty soon ____ clean cleaned up. d. CHI: Thomas need to go. CHI: Because he need to go chug fast FAST Inv: mmhm. CHI: Because my train is fast. CHI: Mister Conductor said ____ won't crashed# he said (9) TOM:

a. CHI: We have to make ____ backwards. b. CHI: Can't fix ____. c. CHI: Needs ____. d. CHI: I found ____.

(10) LEX: a. CHI: I got a wrong ____.

b. CHI: No no no, you gotta do ____ inside the car. c. CHI: Make a big ____.


13

3.3.2 Analysis It becomes immediately clear from the graphs above that TOM and LEX do not pattern with either monolingual English speakers or Carlo. In particular, at Stages 3 and 4 (with the age>5;00 and the MLU>4), the rate of non-adult subject omission in koda English (taken as a mean) hovers above 5%, while both monolinguals and Carlo (a representative of a larger unimodal bilingual population of the relevant type) are well below that. By the same token, while both TOM and LEX exhibit the well-documented (Valian 1991) asymmetry between the rates of subject- vs. object-omission, at least at Stages 3-4, kodas clearly differ from English monolinguals and Carlo in omitting more objects. Two different statistical tests were applied to the data: the individual children’s data were subjected to the Fischer Exact Probability testing, and in cases where the Fisher Exact Test was computationally intractable, the p-value was estimated using a χ2 Test of Association, with Yates' correction for continuity. Additionally, following Serratrice et al. (2004), the difference in rates of argument omission was calculated in z-ratios (two-sample z-test for the difference between proportions, on the assumption that TOM and LEX represent bimodal bilingual populations and Carlo—unimodal bilinguals acquiring the relevant pair of languages).3 The results of the statistical analyses are shown in Tables 8-13. Table 8. NS rate: TOM Stage Comparison with ENGMean Comparison with CarloIt-Eng

1 ptwo-tailed = .0218* [z-] ptwo-tailed = .636 [FET]

ptwo-tailed <.0002* [z-] ------

2 ptwo-tailed = .0005* [z-] ptwo-tailed = .238 [FET]

ptwo-tailed <.0002* [z-] ptwo-tailed = . 002* [FET]

3 ptwo-tailed = .0002* [z-] ptwo-tailed = .16 [χ2]

ptwo-tailed <.0002* [z-] ptwo-tailed = 1 [FET]

4 ptwo-tailed <.0002* [z-] ptwo-tailed <. 0001* [FET]

ptwo-tailed <.0002* [z-] ptwo-tailed =.025* [FET]

Table 9. NS rate: LEX Stage Comparison with ENGMean Comparison with CarloIt-Eng

1 ------ ------ 2 ptwo-tailed <.0002* [z-]

ptwo-tailed<.0001 [χ2] ptwo-tailed <.0002* [z-] ptwo-tailed <. 0001* [FET]

3 ptwo-tailed <.0002*[z-] ptwo-tailed<.0001* [χ2]

ptwo-tailed <.0002* [z-] ptwo-tailed<.0001* [χ2]

4 ptwo-tailed <.0002*[z-] ptwo-tailed <.0002* [z-]

3 One asterisk marking represents full and double-asterisk—marginal significance.


14

ptwo-tailed<.0001* [χ2] ptwo-tailed = .102 [χ2] Table 10. NS rate: KODAMean Stage Comparison with ENGMean Comparison with CarloIt-Eng

1 z= -2.293; ptwo-tailed = .0218* ---- 2 z= 1.579; ptwo-tailed = .1143** z= 7.872; ptwo-tailed <.0002* 3 z= 4.131; ptwo-tailed <.0002* z= 4.618; ptwo-tailed <.0002* 4 z= 5.168; ptwo-tailed <.0002* z= 5.551; ptwo-tailed <.0002*

Table 11. NO rate: TOM Stage Comparison with ENGMean Comparison with CarloIt-Eng

1 ptwo-tailed = .62 [z-] ptwo-tailed = .636 [FET]

----

2 ptwo-tailed =.1964 [z-] ptwo-tailed = .238 [FET]

ptwo-tailed =.0025* [z-] ptwo-tailed = .002* [FET]

3 ptwo-tailed =.1192**[z-] ptwo-tailed =.16 [χ2]

ptwo-tailed =.953 [z-] ptwo-tailed = 1 [FET]

4 ptwo-tailed <.0002*[z-] ptwo-tailed<.0001* [FET]

ptwo-tailed =.0179*[z-] ptwo-tailed = .025* [FET]

Table 12. NO rate: LEX Stage Comparison with ENGMean Comparison with CarloIt-Eng

1 ------ ------ 2 ptwo-tailed =.0096* [z-]

ptwo-tailed = .015* [FET] ptwo-tailed <.0002* [z-] ptwo-tailed <.0001* [FET]

3 ptwo-tailed <.0002* [z-] ptwo-tailed<.0001* [χ2]

ptwo-tailed =.1704 [z-] ptwo-tailed = .248* [χ2]

4 ptwo-tailed =.068* [z-] ptwo-tailed = 1 [FET]

ptwo-tailed =.604 [z-] ptwo-tailed = .73 [FET]

Table 13. NO rate: KODAMean Stage Comparison with ENGMean Comparison with CarloIt-Eng

1 z= -0.49; ptwo-tailed = .62 ----

2 z= 2.114; ptwo-tailed =.0345* z= 3.066; ptwo-tailed =.0022*

3 z= 3.73; ptwo-tailed =.0002* z= 0.7292; ptwo-tailed =.4284

4 z= 3.346; ptwo-tailed =.0008* z= 1.624; ptwo-tailed =.1044**


15

Analysis of verb types reveal that TOM and LEX omit subjects in both transitive and

intransitive contexts, and the vast majority of NO constructions involve verbs that require an overt complement at all times. This is demonstrated in Figures 4-5.

Figure 4. By verb type: NSEng, Eng sessions Figure 5. By verb type: NOEng, Eng sessions In order to establish whether argument omissions are associated with a particular

person/number or verb type (cf. Schmitz et al. 2012), NAs were coded for the contextually applicable grammatical person; this is shown in Figures 6-7.

Figure 6. By person: NSEng, Eng sessions Figure 7. By person: NOEng, Eng sessions As the figures above demonstrate, kodas’ null arguments come in a variety of persons; the vast majority are 1SG and 3SG, the latter particularly prevalent in NO constructions. In the data, no particular verb favored a particular person/number of a missing argument: around 40 different verbs missing arguments were produced by TOM and LEX each, each of the verbs with a variety of persons/numbers. Tense/agreement errors were observed in 27% of 3rd person singular

0% 10% 20% 30% 40% 50% 60% 70% 80% 90%

100%

TOM LEX

Verb types with omitted subjects

intransitive

transitive

0% 10% 20% 30% 40% 50% 60% 70% 80% 90%

100%

TOM LEX

Verb types with omitted objects

allow 'implicit arguments' but not in the context under examination

require complement in all contexts

0%

20%

40%

60%

80%

100%

TOM LEX

Subject omission (person/number)

3 PLU

3rd SG

2nd

1st PLU

1st SG 0%

20%

40%

60%

80%

100%

TOM LEX

Object omission (person/ number)

Other 3 PLU 3rd SG 2nd 1st PLU 1st SG


16

contexts in TOM’s speech and in 11% in LEX’s. Correct suppliance of 3rd person morphology occurred primarily with the verb ‘be’ (as is) but also with lexical verbs such as pull, push, want. Argument omission did not appear to be solely driven by maintaining discourse reference: in the subset of the data analyzed for reference shift, over 50% arguments omitted by TOM and approximately 30% by LEX shifted in reference, which implies that the null argument was not (necessarily) bound by the discourse topic.

Overall, kodas omitted arguments in English at every stage of linguistic development and in a variety of contexts. Of particular relevance are omissions not at the left edge, as subjects of subordinated clauses, and with modals (see (14)-(17)). Such ‘oddities’ from the point of view of developmental English (both monolingual and unimodal bilingual as reported in the literature, cf. Liceras et al., 2012) comprise 25% of TOM’s null subjects and about 18% of LEX’s, and occur at every stage of linguistics development. Figure 8 presents the breakdown by type.

Figure 8. Atypical environments for subject omission in English, English-target sessions

To briefly summarize the results: the subjects of this study do not pattern with either the

monolingual English children or the English-Italian bilingual (Serratrice et al. 2004). The rates and types of null arguments in the koda English (for individual children and as means) differ significantly from those in the English of Sarah, Nina, Naomi, and Adam; moreover, in the relevant respects, kodas perform differently from Carlo—an Italian-English bilingual, serving as a representative of bilinguals acquiring English and a null argument language simultaneously. TOM and LEX omit more and differently then the comparison groups. On the whole, patterns of argument omission in the English of TOM and LEX are fully consistent with argument omission in the children’s other language—ASL (see (4)). Note that the data reported here highlight the path towards answering the first question posed in the previous sections: in the domain of argument omission/suppliance, cross-linguistic interaction does not appear unidirectional, since TOM and LEX’s English, unlike what has been reported in the literature on simultaneous acquisition of similar language combinations, appears

0% 10% 20% 30% 40% 50% 60% 70% 80% 90%

100%

TOM LEX

Unattested Null Subjects

multiple

subject of a question with Aux

subject of embedded clause

with a modal

non-left edge


17

to be affected by ASL. Two questions arise here which warrant further inquiry: whether a) similar effects will be observed in the NA language, as reported in the much of the literature on unimodal bilingualism, and b) argument omission in the English used by the kodas reveals the actual state of their English—that is, whether the rates of argument omission in English remain stable across English- vs. ASL-target sessions; and c) how the uniqueness of bimodal bilinguals can be accommodated under a general theory of language interaction effects in bilingualism? The follow-up study reported below in an attempt to answer (a) and (b) provides a possible direction for the answer to (c). 3.4. Part 2 (follow-up): ASL-target 3.4.1 Data Systematic investigation of argument omission rates in the null argument languages of bilinguals acquiring it concurrently with English has shown that the null argument language becomes affected (see section 2.2). Therefore, we predict a similar effect on ASL, as in (11): (11) The rate of argument omission in the NA-language of ASL-English bilinguals (ASL)

will be lower than what is observed in monolingual ASL production; overt arguments will be supplied in the environments dispreferred by ASL adult grammar = (5b)

Additionally, while the previous sections have shown that TOM and LEX do not (always) obey the constraints on argument suppliance imposed by English, allowing instead ASL-like omissions, the question is whether their performance in ASL mirrors their performance in English.

This part of the study focuses on the ASL-target sessions. As in the English study, utterances were coded based on the presence of a verb; verbs were coded based on whether they required arguments; null arguments were coded for subject- vs. object-omission. Many of the utterances contained items transcribed as gestures and emblems (Lillo-Martin et al. 2012); the former but not the latter were excluded from the analysis.4 Table 14 records the descriptive data.

Table 14. Koda ASL

Subject Stage Nu. of utterances with verbs




Nu. of linguistic utterances

4 In addition, a vast number of utterances were excluded due to ASL-internal reasons: ASL does not have an overt copula. This potentially allows utterances containing an Adjectival predicate only to be viewed as cases of argument omission. (i) a. TOM: IX(self) __ BOY ‘I (am) a boy’ b. LEX: ___ TIRED ‘(I am) tired’ However, in order to remain consistent with the methodology of the previous study, only utterances with an actual verb (excluding imperatives) were included.


18

TOM 3 4

7 20

2 (.30) 10 (.50)

6 6

0 0

200 110

LEX 3 4

113 38

46 (.41) 15 (.40)

41 12

3 (.07) 1 (.08)

406 130

Example utterances are provided below: (12) TOM, bimodal: a. CHI: OUTSIDE ___ GO OUTSIDE NOW I am going outside now b. CHI: g(no)[+] ___ CLIMB IX(self) NEVER FALL Ø CLIMB I never fall when I climb (13) LEX, unimodal:

a. CHI: BECAUSE ____ WANT IX (toy) ‘Because (I) want that’

b. CHI: IX(self) HAVE ____ ‘I have (that)’ Results are represented graphically in comparison with Carlo in Figures 9-10.

Figure 9. NSASL, ASL sessions Figure 10. NOASL, ASL sessions

Two ASL consultants found all cases with argument omission acceptable in the adult language. Among the overt arguments supplied by the children, two were considered inappropriate by one adult Deaf consultant – a ‘redundant’ use of 1-IX, typically glossed as the 1st SG pronoun.

Additionally, since a number of works have shown that bimodal bilingual children typically use both languages simultaneously (cf. Lillo-Martin et al., 2009; et seq.), English used during the ASL-target sessions was examined independently (Table 15).

0 0 7 8 1 0

0

20

40

60

80

100

Stage 3 Stage 4

% o

f NO

in th

e N

A la

ngua

ge

Object omission (individual children): NA language

TOM

LEX

Carlo 30

50 41 40

59 56

0

20

40

60

80

100

Stage 3 Stage 4

% o

f NS

in th

e N

A la

ngua

ge

Subject omission (individual children): NA language

TOM

LEX

Carlo


19

Table 15. Koda English, sorted by stages, ASL-target sessions Subject Stage Nu. of

utterances with verbs





TOM 3 4

91 86

10 (.12) 18 (.21)

84 62

9 (.11) 1 (.02)

345 274

LEX 3 4

245 74

25 (.10) 1 (.01)

233 58

4 (.02) 2 (.03)

326 303

The rates of subject and object omission in the children’s English during the ASL-target sessions are plotted in Figures 11-125; (14)-(15) record examples of utterances with omitted arguments.

Figure 11. NSEng, ASL-target sessions Figure 12: NOEng, ASL-target sessions (14) TOM, unimodal:

a. CHI: because ___ want that b. CHI: ___ hid the house c. CHI: Put them ___

(15) LEX, unimodal: a. CHI: ___ turn and sit just his butt b. CHI: ___ think come back here, you tires c. CHI: Ø don’t know around ___

3.4.2. Analysis

5 Since Serratrice et al. (2004) do not report amount of English uttered during the Italian-target sessions, Carlo’s rates of argument omission in English are not recorded here.

0

20

40

60

80

100

Stage 3 Stage 4 % o

f NS

in k

oda

Engl

ish

Subject omission (individual children): English

TOM LEX 11

2 3 0

20

40

60

80

100

Stage 3 Stage 4 % o

f NO

in k

oda

Engl

ish

Object omission (individual children): English

TOM LEX


20

Overall, in their NA language (ASL), both kodas omit subjects and objects. During Stages 3-4, kodas omit fewer subjects and more objects than Carlo. At the same time, while kodas’ ASL exhibits the subject-object asymmetry ordinarily noted in the literature (including that on monolingual ASL acquisition), the rates of argument omission do not mirror monolinguals’ performance as adults (Wulf et al. 2002) or young children (Lillo-Martin 1991). At Stage 4 (at the age≥5), kodas’ subject omission hovers around 40% and object omission at 5% -- clearly less than in the adult language. However, drawing any conclusions from the rate itself is dangerous: since all of the cases of argument omission are compatible with adult grammar, and only in two cases of overt argument suppliance (both in reference to self, see (4a)) did native ASL consultants disagree, the answer to the question whether kodas behave like monolingual children at these stages must await until the comparable data from monolinguals are available (but see Reynolds 2015).

It is clear, however, that TOM and LEX differentiate between their two languages. In both ASL and English, the children omit subjects and objects, but they do so differently depending on the language: the rates of argument omission in ASL are always higher than in English.6 This is evident from the Tables 16-17, where the data are collapsed across the two children and the mean (KODAMean) is provided for each stage.

Table 16. KODAMean ASL, sorted by stages






3 4

120 58

48 (.40) 25 (.43)

47 18

3 (.06) 1 (.05)

606 240

Table 17. KODAMean English, sorted by stages






3 4

336 160

35 (.10) 19 (.12)

317 120

13 (.04) 3 (.009)

671 577

As Table 18 demonstrates, the difference between the rates of subject (though not object)

omission in ASL vs. English is significant (reported in z-ratios). Table 18. KODAMean NA rate in ASL vs. English: ASL-target sessions Stage Null subject Null Objects

6 A caveat is in order: in the transcripts at Stage 3, TOM does not omit any arguments in ASL; however, this may very well be a function of the choice of predicates, none of which actually require overt objects: PLAY, EAT, SLEEP, i.a. More data are necessary here.


21

3 z= 7.2; ptwo-tailed < .0002* z= .712; ptwo-tailed =.4765 4 z= 5.76; ptwo-tailed < .0002* z= .712;ptwo-tailed = .4709

In other words, as the table above attests, TOM and LEX are differentiating between their languages. While this is not statistically confirmed for the object omission – which may be due to the ‘floor effect’ – it is undoubtedly so for the subject omission: in the same set of sessions, kodas, using both languages, omit subjects on average around four times more often in ASL than they do in English. 4. Discussion 4.1 Overview of the results The findings of the study reported above demonstrate the following: ASL-English bilinguals TOM and LEX pattern with English monolinguals with respect to overt argument suppliance at the only earlier stages of linguistic development only, and only in terms of numbers—i.e. at the later stages, they perform differently from either monolingual or bilingual English comparisons. Crucially for the predictions outlined in (5), at Stages 3 and 4 (the latter capturing the linguistic development of the children at almost 5 years of age), TOM and LEX continue to omit arguments in contexts disallowed in adult English. Individually and in terms of mean, omission of subjects and objects is significantly different from monolinguals and from Carlo. In all of the aforementioned cases, however, ASL allows arguments to be silent; thus, the most logical step to take is to admit cross-linguistic interaction effects. That is, as the study above demonstrates, it is not the case that the unidirectionality effects related to the overt argument suppliance observed in many language pairs must be built into the model of cross-language interaction.

Particularly striking is the fact that among such contexts is the embedded subject position of a finite clause, the subject of a question, and a non-1st person ‘diary-drop’ – never a grammatical option in English.

(16) a. *Shei/j said ___ i/j will come.

a. *Where did____ go. b. ___1SG/*3SG Gotta see a doctor tomorrow.

Yet, some of the examples in (7)-(8), consolidated below, reveal NSs in exactly this configuration: (17) a. CHI: It says ___ hafta build a king of the king’s horsies. Target: It says I have to build a king (or a king’s house?) for the king’s horses b. CHI: Mister Conductor said ___won't crashed# he said. Target: Mister Conductor said that that it wouldn’t crash; that’s what he said

b. CHI: How did ___ tape it? Target: How did you tape it?


22

Note that these occur at the stage of linguistic development associated with morpho-syntactic (and not discourse, cf. Rizzi 2005; i.a.) licensing in the language of children who are already attending school—i.e. whose language can be considered having reached a ‘developed’ stage. Although accounts of null subjects in child English vary, none of them predicts (17) in general and at this stage in particular (see section 2.1).

A similar observation can be made about the NO constructions recorded in (9)-(10), repeated here as (18):

(18) a. CHI: We have to make __ backwards.

Target: we have to make this/it backwards b. CHI: Can't fix ___ .

Target: I can’t fix it. c. CHI: Needs Ø.

Target: He[?] needs me[?] d. CHI: I found Ø.

Target: I found you. e. CHI: No no no, you gotta do Ø inside the car.

Target: No no no, you gotta do this/it inside the car.

Although English allows some object omission (as in (2)), utterances produced by the children in this study and coded as missing objects are never grammatical in adult, or found in child, English (Snyder et al. 2001). In other words, the type of NS in (17) and NO in (18) in the English of simultaneous ASL-English bilinguals is unexpected and thus constitutes (for our purposes) the first puzzle to be solved. To be precise, the English sentences above seem to allow the NA in a manner consistent with the grammatical properties of ASL, consistent with the possibility of ‘transfer’ from the NA language into the non-NA language. The question then is ‘transfer’ of what?

Let us then take the leap. Suppose the NA in the TOM and LEX’s English does, in fact, come from ASL. In that case, a caveat is in order: is not entirely clear whether a comparison between TOM and LEX vs. Carlo is appropriate. While some (e.g. Bahan et al., 2000) have argued that the NA in ASL is identical to that in Italian—i.e. it is pro,—others disagree. For instance, Lillo-Martin (1991), i.a., maintains that pro in ASL occurs with agreeing verbs only, and the NA is a deleted topic otherwise; Author (2012) offers an NP ellipsis account instead. Incidentally, more recently, the discussion of Italian pro has expanded: Frascarelli (2007, i.a.) has argued that in addition to being licensed by clause-internal mechanisms, pro must also be bound by the aboutness topic. For the purposes of this paper, this move unifies Bahan et al. (2000) and Lillo-Martin (1991) accounts and predicts the following: if the NA in koda’s English is, in fact, the relevant element from ASL, we expect it to be conindexed with the NP which happens to be the topic being maintained. Yet, as the data show, this is not always the case (see section 3.3.1): while the analysis of reference shift in ASL is currently impossible for external


23

reasons, a comment is warranted: if whatever allows the NA in ASL somehow shows up in the English of kodas in this study, we might say that nether Bahan et al. nor Lillo-Martin can explain the fact that a fair amount, but not all, of argument omission cases occur with shifted reference.

However, one cannot simply say that the NAs in the children’s English transfer directly from ASL—i.e. that ASL-English bilinguals, mistakenly, though perhaps temporarily, adopt some property of ASL into their English. As the second part of the study demonstrates, although kodas omit arguments in both languages, they do that to a much higher degree in ASL. This warrants a comparison: Figures 13-16 below show the rates of argument omission at Stages 3-4 per language (where available) per language target; the target language of the session is capitalized, the language under examination is not.

Figure 13. NSASL / English, ASL vs. English sessions, individual children

Figure 14. NOASL / English, ASL vs. English sessions, individual children

0

20

40

60

80

100

TOM ASL_asl

TOM ASL_eng

TOM ENG_eng

LEX ASL asl LEX ASL_eng

LEX ENG_eng

% o

f sub

ject

om

issi

on

Subject omission (individual children): ASL and English

Stage 3 Stage 4

0

20

40

60

80

100

TOM ASL_asl

TOM ASL_ eng

TOM ENG_eng

LEX ASL_asl LEX ASL_eng

LEX EN_ eng

% o

f obj

ect o

mis

sion

Object omission (individual children): ASL and English

Stage 3 Stage 4


24

Figure 15. NSASL/Eng, ASL vs. Eng sessions Figure 16. NOASL/English, ASL vs. Eng sessions As the graphs above illustrate, the rate of subject omission differs between ASL and English; this difference is statistically significant. Moreover, at Stage 4, the difference in rates of argument omission in English between the ASL- and English-target sessions is also significant (Table 19).

Table 19. KODAMean argument omission in English: ASL- vs. English-target sessions Stage Null subject Null Objects

3 z= .0706; ptwo-tailed < .474 z=.88; ptwo-tailed =.379 4 z=1.838; ptwo-tailed < .068* z= 2.41; ptwo-tailed = .016*

The data above lead to two conclusions: although ASL-English bilinguals in the studies reported here use both languages consistently, irrespective of the pragmatic appropriateness of one vs. the other, they clearly differentiate between the two languages. That is, their ASL differs in the relevant respects from their English. At the same time, the children—especially at later ages—appear aware that the English-target discourse necessitates English, while the ASL-target discourse requires ASL: the rates of ASL-like properties of koda English differ based on the target language of the session—more ASL-like English is more appropriate in ASL-target sessions than in the English-target sessions. In other words, while the languages of kodas appear to be interacting, the ‘grammars’ remain differentiated. This finding, incidentally, lends independent support to the claim that bilinguals keep certain aspects of their grammars separate. However, it also presents us with another set of puzzles: a) why should any English be used during the ASL session?, and b) why do ASL-English bilinguals stand alone as “victims” of what might be termed a null argument- onto a non-null argument language influence? I would like to suggest the answers to both of these questions are rooted in the same phenomenon—an ability to use two languages simultaneously.

40

10 12

43

12 8

0

20

40

60

80

100

KODA ASL_asl

KODA ASL_eng

KODA ENG_eng

% o

f sub

ject

om

issi

on

Subject omission (means): ASL and English

Stage 3

Stage 4

6 4 5 5 0.9 4 0

20

40

60

80

100

KODA ASL_asl

KODA ASL_eng

KODA ENG_eng

% o

f obj

ect o

fmis

sion

Object omission (means): ASL and English

Stage 3

Stage 4


25

4.2. Bimodality Using converging methodologies, Kroll and colleagues (Kroll et al., 2006; see also Costa, Miozzo & Caramazza, 1999; i.a.) have argued that in the bilingual mind, both languages are always active, and one of them needs to be inhibited for the purposes of production of the other. Arguably, this task demands a certain amount of mental resources; thus, controlled ‘juggling’ of more than one language places a unique cognitive burden on bilinguals. In various works, Sorace (see an overview in Sorace, 2011) has suggested that in this scenario, overt arguments appear. These elements are not ‘grammaticalized’ (Asudeh, 2004); they might be thought of as ‘complexity resumptives’ (Erteschik-Shir, 1992), which, albeit unnecessary from the point of view of syntax, arise due to processing complexity, as in (19): (19) This is the girl that Peter said that John thinks that yesterday his mother had given

some cakes to (her). (Erteschik-Shir,1992) Crucially, this effect is expected irrespective of (and despite) the language requirements for overt argument suppliance. The outcome of this view is that what appears on the surface to be a case of unidirectional influence from a non-NA language (like English) into a NA language (like Italian or Japanese) may in reality result from some other bilingual effect associated with inhibiting one of the languages. Thus, even if language-specific interaction, resulting in argument omission in the non-NA language, were possible, its effects on argument omission/suppliance rates would be obscured by the presence of other (e.g. processing related) factors.

The aforementioned implies that the high rate of overt arguments in Italian of an Italian-English bilingual does not betray ‘transfer’ from English. Rather it reflects a general bilingualism effect (i.e. a characteristic of a linguistic mind that is unique to bilinguals) and is, then, also predicted to surface in the languages of a bilingual if both of her languages allow NAs, e.g. Spanish-Italian bilinguals. This prediction appears supported empirically (see section 1.2.1). However, arguably, it also constitutes a unimodal bilingualism effect.

In a number of studies, Emmorey and colleagues (Emmorey, Borinstein, Thompson & Gollan, 2008; Emmorey, Luk, Pyers & Bialystok, 2008; Emmorey & McCullough, 2008; Emmorey & McCullough, 2009; Giezen et al., 2015) have argued that what makes a bimodal bilingual different from a unimodal bilingual is simultaneous access to two modes of conveying a message. The authors suggest that a bimodal bilingual uses the gestural system, independently available to all (unimpaired) language users, in producing an utterance in both languages simultaneously. Emmorey et al. (2008) represent the architecture of the bimodal bilingual brain as in Figure 17.


26

Figure 17. Bimodal bilingual language architecture (Emmorey et al. 2008)

This view of the processes associated with bilingual production allows a bimodal bilingual to encode the message in both languages (code-blend, Emmorey et al., 2008; Casey & Emmorey et 2009, i.a.). Note a particular consequence of the model in Figure 17: what is possible is simultaneous selection of two lexical representations but not two propositions – although parts of the utterance can come from two different languages, they constitute the same proposition (see also Donati & Branchini 2009; et seq.). This, in turn, translates into the following: in bimodal bilingual production, neither language needs to be inhibited. The aforementioned predicts that bimodal bilinguals will code-switch (i.e. use alternate overt lexical items from both of their languages) less, as compared to unimodal bilinguals; instead, they will use both languages simultaneously – i.e. code-blend. And, in fact, this prediction is supported empirically. Studies have shown that bimodal bilinguals exhibit a strong preference for code blending (90%) over code switching (<10%) (Emmorey et al., 2008; van den Bogaerde & Baker, 2005; Petitto et al., 2001).

Emmorey et al. conclude that lexical suppression must be more costly computationally than lexical selection. The flip side of this observation relates to the presence (or lack thereof) of the ‘bilingualism advantage’ in executive control. As research has shown, the constant practice in having to choose the right language for the context (or rather, to inhibit the other one) leads to a bilinguals’ tendency to outperform monolinguals on tests involving competing stimuli (see an overview in Bialystok, 2009; i.a.). However, since an ASL-English bilingual, unlike, e.g., an Italian-English bilingual, is not constantly required to inhibit one of the languages, this kind of ‘bilingualism advantage’ is unexpected. As Emmorey, Luk, Pyers and Bialystok (2008) show, this prediction is borne out as well: ASL-English bilinguals perform with monolinguals, and not with unimodal bilinguals, on the tasks related to sorting out competing stimuli (e.g. Flanker and picture-naming tasks). This ability of bimodal bilinguals to produce code-blends without any cost to either of the languages implies a potential to bypass the competition.

The view of bimodal bilingualism outlined above suggests that the examination of the linguistic patterns of this population makes it possible to hold the general bilingualism effects


27

constant. If language users are not required to (necessarily) inhibit (the production of) one of their languages, then the processing difficulties caused by tasks such as anaphora resolution are diminished.7 In such a scenario, language-specific interaction effects will be allowed to surface. Therefore, if the theory of bilingualism effects requires a model of language-specific interactions, independent from the general bilingualism effects, this need will rise to the surface with bimodal bilinguals. 5. Account and directions The path just outlined brings us to a surprising revelation: if ASL-English bilinguals in the studies reported here perform differently from unimodal bilinguals due to the lack of forced language choice, then it stands to reason that the field should look to this this population for empirical observations regarding bilingual language comprehension/production—in this population, language interaction effects are not masked. The question then is what is behind these language interaction effects. A number of proposals are found in the literature, an influential one among which is the cross-linguistic influence by Hulk & Müller (2000, i.a; see fn. 2). The model predicts language interaction effects in the domains where the input is ambiguous (Müller 1998), suggesting that young bilingual children (crucially, prior to the full development of morpho-syntactic licensing mechanisms) may temporarily adopt an analysis for one of their languages that is appropriate only for the other. Various works have argued against this approach to bilingual language interaction effects, in the account of bimodal bilingual production in particular (cf. Lillo-Martin et al., 2009; et. seq.; for a detailed discussion, see Author 2012). Instead, I assume the view that bilingual linguistic system can be represented as in Figure 18 – via Minimalist-style code-switching a là MacSwan (2000):

7 A prediction arises here: in anaphora resolution tasks, bimodal bilinguals should differ from unimodal bilinguals in allowing argument ellipsis in anaphora resolution tasks in their NA language. Author (2013a) has demonstrated that kodas behave as predicted.


28

Figure 18. Bilingual architecture (Lillo-Martin et al., 2009, Author, 2012): Language synthesis This view implies that the linguistic system of a bilingual contains lexical items from both languages (Lx and Ly in Figure 18). At the point of vocabulary insertion, elements from either language can be inserted, as long as PF requirements are satisfied; otherwise, code-switch is banned precisely because the derivation crashes (see Cantone, 2007 for an account of unimodal code-mixing in child unimodal bilingualism along the same lines). Here is the upshot of such an approach with regards to argument omission: the NA in a language like ASL has been argued to result from the type of T° and v° in the language (with vs. without φ-features: Holmberg, 2005; Saito, 2007; Biberauer et al., 2010; i.a.). In conjunction with the view of code-switching outlined above, an ASL-English derivation in Figure 19 becomes possible (see González-Vilbazo & Lòpez, 2011, i.a. for various other language combinations: German-Spanish, Taiwanese-English, etc).

Figure 19. A possible ASL-English CP (after González-Vilbazo & Lòpez, 2011) The model in Figure 18, resulting in the tree in Figure 19, offers an appealing account of language interaction effects that might otherwise be labeled ‘transfer’/‘cross-linguistic influence’: the presence of the ASL T° and v° in the otherwise English clause will result in the possibility of argument omission—i.e. the clause that may appear language uniform may, in fact, contain phonologically null elements (e.g. functional heads) from another language (see Author 2012 for an extensive discussion). Note that the aforementioned is entirely divorced from the knowledge (or lack thereof) of the morpho-syntactic processes in both of the languages of the bilingual or their misanalysis (pace, e.g. Hulk and Müller), as it is from the rates of code-blending, unless two concurrent derivations is a possibility (see Donati & Branchini, 2009 for a discussion). Instead, what is at stake is accessing the appropriate set of lexical items for the Lexical Array—i.e. as he approaches sentence production/comprehension, what the bilingual child needs to learn is which language to ‘pull from.’ Arguably, this constitutes a separate type of


29

learning.8 This stance further suggests that in the context where such code-switch is particularly appropriate—such as during an interaction with other bilinguals with the same/similar language combination—the rates of phenomena of this sort will be elevated. Translating the aforementioned into a general description of bilingual language production: in an environment that naturally lends itself to code-switching, ‘cross-linguistic influence’ will be observable at higher rates. In other words, the theory of language interaction can be described in terms of a theory of language choice. Adding to this observation the fact that bimodal (e.g. ASL-English) bilinguals are much less practiced in inhibiting one of their languages than their unimodal (e.g. Italian-English) counterparts results in a potential possibility that bimodal bilinguals will exhibit effects of language synthesis to a higher degree, in more domains, and for longer than unimodal bilinguals. This view seems to be supported by various empirical findings, the studies reported here included. 6. Conclusion

This paper has reported two studies of argument omission in spontaneous production of ASL-English bilinguals: both in ASL and English. The study was chiefly guided by the question whether ASL-English bilinguals would exhibit cross-linguistic effects in argument omission in each of their languages—arguably, the consequence of their other language. We have seen that ASL-English bilinguals exhibit elevated/protracted rates of argument omission in their English typically unobserved in monolinguals or unimodal bilinguals; the children’s rate of argument omission in ASL is also lower than what is reported in the literature. It is, of course, possible that TOM and LEX are not the norm. The concern then is this: if it is impossible for children of this demographic to omit arguments (for whatever reason), TOM and LEX, while being outliers, show that such ‘impossibility’ does not subsume bimodal bilinguals. Further examination of monolingual and unimodal bilingual signers promises to shed further light on how different koda ASL actually is. We have also shown that the children clearly differentiate between their two languages: in the ASL-target sessions, they used both English and ASL; however, the rates of argument omission in English differ from that in ASL. Further, at the latest stage of linguistic development, the rates of argument omission in koda’s English during ASL-target sessions differ from those in the English-target sessions. I interpret this finding as suggesting of the fact that the children are improving in accessing the language-appropriate set of lexical items. The follow-up study supports the view advocated in Emmorey et al. (2008, i.a.) and Lillo-Martin et al. (2010, i.a.) that bimodal bilinguals use both languages simply because they can: the children in the study consistently use both languages simultaneously. On the approach to bilingual language production in Sorace (2011, i.a.), a prediction arises that in anaphora

8 DenDikken (2011) argues that functional elements from multiple langauges actually compete in the linguistic mind of the bilingual. On this view, the linguistic mind of a bilingual is affected by the knowledge of more than one language is a non-trivial manner: language-synthesis becomes obligatory and interesting predictions arise.


30

resolution tasks, bimodal bilinguals will behave differently from unimodal bilinguals as well— they are not expected to oversupply overt arguments in their ASL, at least for processing reasons (see Author, 2013a, for confirmation). Note that if what passes for language interaction effects between the languages of a bilingual results from a forced language choice, then an ability to bypass this choice (at least to some degree) may yield different outcomes. This line of reasoning interfaces with our account of the language interaction effects regarding null arguments observed, to date, only in ASL-English bilinguals. We argue here that the data point to the direction of an explanation along the lines of code-switching (or, as it has been termed in the recent literature, language synthesis), and not dominance or cross-linguistic influence. Many more questions remain, however.

References: Allen, S. (2000). A discourse-pragmatic explanation for argument representation in child

Inuktitut. Linguistics, 38(3), 483-521. Allen, S. (2001). The importance of discourse-pragmatics in acquisition. Bilingualism:

Language and Cognition, 4(1), 23-25. AnderBois, S. (2011). Issues and Alternatives. Unpublished doctoral dissertation. University of

California, Santa Cruz. Asudeh, A. (2004). Resumption as Resource Management. Unpublished doctoral dissertation,

Stanford University. Bahan, B. (1996). Non-manual Realization of Agreement in American Sign Language,

Unpublished doctoral dissertation, Boston University. Bhatt, R. & Pancheva, R. (2006). Implicit arguments. Blackwell Companion to Syntax, 558-588.

Oxford: Blackwell. Bialystok, E., Craik, F., Green, D., & Gollan, T. (2009). Bilingual Minds, 10(3), 89–129. Biberauer, T., Holdmberg, A., Roberts, I., & Sheehan, M. (2010). Parametric Variation: Null

Subjects in Minimalist Theory. Cambridge: Cambridge University Press. Bishop, M. (2009). Bimodal bilingualism in hearing, native users of American Sign Language.

Unpublished doctoral dissertation, Gallaudet University. Blais, M.J., Oshima-Takane, Y. Genesee, F. & Hirakawa, M. (2010). Cross-linguistic influence

on argument realization in Japanese-French bilinguals. Proceedings of the 34 annual Boston University Conference on Language Development (pp. 34-45). Somerville, MA: Cascadilla Press.

Bloom, P. (1990). Subjectless sentences in child language. Linguistic Inquiry 2(1), 491-504. Borer, H. & Wexler, K. (1987). The maturation of syntax. In T. Roeper &Williams, E., (Eds.)

Parameter Setting, 123-172. Dordrecht: Reidel,. Borer, H., & Rohrbacher, B. (2002). Minding the absent: Arguments for the full competence

hypothesis. Language Acquisition 10(2), 123-175. Boster, C. T. (1997). Processing and parameter setting in language acquisition: A computational

approach. Unpublished doctoral dissertation, University of Connecticut, Storrs.


31

Brackenbury, T., Ryan, T., & Messenheimer, T. (2006). Incidental word learning in a hearing child of deaf adults. Journal of Deaf Studies and Deaf Education, 11(1), 76-93.

Brown, R. (1973). A First Language. Cambridge, MA: Harvard University Press Cantone, K., & Müller, N. (2005). Codeswitching at the interface of language-specific lexicons

and the computational system. International Journal of Bilingualism 9(2), 205-225. Cantone, K. F. (2007). Code-switching in Bilingual Children. The Netherlands: Springer Capirci, O., Iverson, J. M., Montanari, S., & Volterra, V. (2002). Gestural, signed and spoken

modalities in early language development: The role of linguistic input. Bilingualism: Language and Cognition 5(1), 25-37.

Casey, S., & Emmorey, K. (2009). Co-speech gesture in bimodal bilinguals. Language and Cognitive Processes 24(2), 290-312.

Chamberlain, C., & Mayberry, R. I. (2008). American Sign Language syntactic and narrative comprehension in skilled and less skilled readers: Bilingual and bimodal evidence for the linguistic basis of reading. Applied Psycholinguistics 29 (3), 367-388.

Chen Pichler, D., Hochgesang, J., Lillo-Martin, D. & Müller de Quadros, R. (2010). Conventions for sign and speech transcription in child bimodal bilingual corpora. Language, Interaction and Acquisition/Language, Interaction et Acquisition 1(1), 11-40.

Costa, A., Miozzo, M., & Carmazza, A. (1999). Lexical selection in bilinguals: Do words in the bilingual’s two lexicons compete for selection? Journal of Memory and Language. 41, 365–397.

DenDikken, M. (2011). The Distributed Morphology of code-switching. Paper presented at 2010 UIC Bilingualism Forum. University of Chicago, IL.

Donati,C. & Branchnini, C. (2009). Challenging linearization: Simultaneous mixing in the production of bimodal bilinguals. Proceedings of the European Summer School on Language, Logic and Information, Bordeaux, France, July 2009.

Emmorey, K., Borinstein, H. B., Thompson, R., & Gollan, T. H. (2008). Bimodal bilingualism. Bilingualism: Language and Cognition 11(1), 43-61.

Emmorey, K., & McCullough, S. (2009). The bimodal bilingual brain: Effects of sign language experience. Brain and Language 109(2-3), 124-132.

Emmorey, K., Luk, G., Pyers, J.E., & Bialystok, E. (2008). The source of enhanced cognitive control in bilinguals. Psychological Science 19(12), 1201-1206.

Erteschik-Shir, N. (1992) Resumptive pronouns in islands. In Goodluck, H. &. Rochemont, M. (eds.) Island Constraints: Theory, Acquisition and Processing (pp. 89-108). Dordrecht: Kluwer,

Frascarelli, M. (2007). Subjects, topics and the interpretation of pro: A new approach to the Null Subject Parameter. Natural Language and Linguistic Theory 25, 691–734.

Gerken, L. (1991). The metrical basis for children’s subjectless sentences. Journal of Memory and Language 30(4), 431-451.

Giezen, M. R., Emmorey, K., & Blumenfeld, H. K. (2015). Parallel language activation and inhibitory control in bimodal bilinguals. Cognition, 141, 9–25.


32

González-Vilbazo, K. & Lòpez, L. (2012). Little v and parametric variation. Natural Language

and Linguistic Theory 30(1), 30-77. Hacohen, A., & Schaeffer, J. (2007). Subject realization in early Hebrew/English bilingual

acquisition: The role of crosslinguistic influence. Bilingualism: Language and Cognition 10(3), 333-344.

Guilfoyle, E. 1984. The acquisition of tense and the emergence of lexical subjects. McGill Working Papers in Linguistics 2(1), 20-30.

Haznedar, B. (2010). Transfer at the syntax-pragmatics interface: Pronominal subjects in bilingual Turkish. Second Language Research 26(3), 355–378.

Hoekstra, T., & Hyams, N. (1998). Aspects of root infinitives. Lingua 106, 91–112. Holmberg, A. (2005). Is There a Little Pro? Evidence from Finnish. Linguistic Inquiry 36, 533-

564. Huang, P-Y. (1999). The Development of Null Arguments in a Cantonese-English Bilingual

Child. Unpublished Masters in Philology thesis, Chinese University of Hong Kong. Hughes, M. & Allen, S. (2008). Child-directed speech and the development of referential

choice in child English. Presented at the International Association for the Study of Child Language Conference, Edinburgh, Scotland, July 2008.

Hulk, A., & Müller, N. (2000). Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition 3(3), 227-244.

Hyams, N. (1986). Language Acquisition and the Theory of Parameters. Dordrecht: Reidel. Hyams, N. (2011). Missing Subjects in Early child Language. In J. De Villiers and T. Roeper

(eds.), Handbook of Generative Approaches to Language Acquisition, 13-52. The Netherlands: Springer.

Hyams, N. and Wexler, K. (1993). On the grammatical basis of null subjects in child language. Linguistic Inquiry 24(3), 421-459.

Jarque, M. J. (2010). Sign bilingualism: Language development, interaction, and maintenance in sign language contact situations. International Journal of Bilingual Education and Bilingualism 13(2), 265-268.

Johnson, J., Watkins, R. & Rice, M. (1992). Bimodal bilingual language development in a hearing child of deaf parents. Applied Psycholinguistics 13(1), 31-52.

Juan-Garau, M. & Perez-Vidal, C. 2000. Subject realization in the syntactic development of a bilingual child. Bilingualism: Language and Cognition 3, 173–192.

Kovelman, I., Baker, S.A., & Petitto, L.A. (2008). Bilingual and Monolingual brains compared: An fMRI investigation of syntactic processing and a possible "neural signature" of bilingualism. Journal of Cognitive Neuroscience 20(1), 153-169.

Kovelman, I., Shalinsky, M. H., White, K. S., Schmitt, S. N., Berens, M. S., Paymer, N., & Petitto, L. (2009). Dual language use in sign-speech bimodal bilinguals: FNIRS brain-imaging evidence. Brain and Language 109(2-3), 112-123.


33

Author (2012). When the quiet surfaces: ‘Transfer’ of argument omission in the speech of ASL-English bilinguals. Unpublished doctoral dissertation, University of Connecticut.

Author (2013a). She said __ was ok: Embedded subject omission in the English of ASL-English bilinguals. Presented at the LSA 2013 (Winter). Boston, MA, January 2013.

Kroll, J. F., Bobb, S. C., & Wodniecka, Z. (2006). Language selectivity is the exception, not the rule: Arguments against a fixed locus of language selection in bilingual speech. Bilingualism 9(2), 119-135.

Lillo-Martin, D. (1991). Universal Grammar and American Sign Language: Setting the Null Argument Parameters. Dordrecht: Kluwer.

Berk, Stephanie & Lillo-Martin, Diane (2012). The two-word stage: Motivated by linguistic or cognitive constraints? Cognitive Psychology 65, 118-140.

Lillo-Martin, D., Quadros, R., Koulidobrova, H. & Chen Pichler, D. (2009). Bimodal Bilingual Cross-Language Influence in Unexpected Domains. Proceedings of Generative Approaches to Language Acquisition (GALA), Lisbon, Portugal, September 2009.

Lillo-Martin, D., Koulidobrova, H., Quadros, R. Müller de & Chen Pichler, D. (2010). Bilingual Language Synthesis: Evidence from WH-Questions in Bimodal Bilinguals. Proceedings of the 36th Annual Boston University Conference on Language Development. Somerville, MA: Cascadilla Press.

MacSwan, J. (2000). The architecture of the bilingual language faculty: Evidence from code-switching. Bilingualism: Language and Cognition 3(1), 7-54.

MacWhinney, B. (2004). The CHILDES Project: Tools for Analyzing Talk. 3rd Edition. Mahwah, NJ: Lawrence Erlbaum Associates.

Marshall, J., Atkinson, J., Woll, B., & Thacker, A. (2005). Aphasia in a bilingual user of British Sign Language and English: Effects of cross-linguistic cues. Cognitive Neuropsychology 22(6), 719-736.

Messing, L.S. (1999). Two modes – two languages? In L. S. Messing & R. Campbell (eds.) Gesture, Speech, and Sign, 83-200. Oxford University Press, New York.

Mishina-Mori, S. 2007. Argument representation in Japanese/English simultaneous bilinguals: Is there a crosslingusitic influence? Proceedings of BUCLD 31, 441- 450. Somerville, MA: Cascadilla Press.

Morgan, G. (2000). Discourse cohesion in sign and speech. International Journal of Bilingualism 4(3), 279-300.

Müller, N. (1998). Transfer in bilingual first language acquisition. Bilingualism: Language and Cognition 1(3), 151-175.

Murphy, J., & Slorach, N.(1983). The language development of pre-preschool hearing children of deaf parents. British Journal of Disorders of Communication 18(2), 118-126.

Petitto, L.,A., Katerlos, M., Levy, B.G., Gauna, K., Tetreault, K., & Ferraro, V. (2001). Bilingual signed and spoken language acquisition from birth: implications for the mechanisms underlying early bilingual language acquisition. Journal of Child Language 28, 453-496.


34

Petitto, L. A., & Holowka, S. (2002). Evaluating attributions of delay and confusion in young bilinguals: Special insights from infants acquiring a signed and spoken language. Sign Language Studies 3(1), 4-33.

Petitto, L. A. & Kovelman, I. (2003). The Bilingual Paradox: How signing-speaking bilingual children help us to resolve it and teach us about the brain’s mechanisms underlying all language acquisition. Learning Languages 8(3), 5-18

Pinto, M. (2013). Tracking reference with null subjects. Linguistics in the Netherlands, 30(1), 131–145. Pizer, G. (2008). Sign and Speech in Family Interaction: Language Choices of Deaf Parents and

their Hearing Children. Unpublished doctoral dissertation, University of Texas, Austin. Quadros, R. M. d., Lillo-Martin, D., & Mathur, G. (2001). What does deaf children's language

acquisition teach us about the stage of optional infinitives? Letras De Hoje 36(3), 391-397.

Quadros, Ronice Müller de & Lillo-Martin, Diane (2007). Aquisição das Línguas de Sinais e a morfologia verbal nas Línguas de Sinais Brasileira e Americana. In Anais do I Encontro do Nordeste em Aquisição da Linguagem (I ENEAL 2005) 1, 1-14. Recife: UNICAP (CD ROM).

Reynolds, W. (2015). Reference tracking in the narratives of early bimodal bilingual children. Presented at Theoretical Issues in Sign Language Research (TISLR) 12, Melbourne, Australia, January 2016.

Rizzi, L. (1986). Null objects in Italian and the theory of pro. Linguistic Inquiry 17(3), 501–557. Roberts, I. & Holmberg, A. (2010). Introduction: Parameters in minimalist theory. In T.

Biberauer, A. Holmberg, I. Roberts & M. Sheehan (Eds.), Parametric variation: Null subjects in minimalist theory (pp. 1-57). Cambridge: Cambridge University Press.

Roeper, T. and Weissenborn, J. (1990). How to make parameters work: Comments on Valian. In Frazier, L. and de Villiers, J. (Eds.) Language Processing and Language Acquisition. (pp. 147-162). Dordrecht: Kluwer.

Sachs, J., Bard, B. & Johnson, M. (1981). Language learning with restricted input: Case studies of two hearing children of deaf parents. Applied Psycholinguistics 2 (1), 33-54.

Saito, M. (2007). Notes on East Asian Argument Ellipsis. Language Research 43, 203-227. Sano, T. and N. Hyams (1994) Agreement, Finiteness, and the Development of Null

Arguments.Proceedings of NELS 24, 544-558. Schiff, N., & Ventry, I. (1976). Communication problems in hearing children of deaf parents.

Journal of Speech & Hearing Disorders, 41(3), 348-358. Schiff-Myers, N. (1988). Hearing children of deaf parents. In D. Bishop and K. Mogford (Eds.)

Language Development in Exceptional Circumstances (pp. 47-61). Edinburgh etc.: Churchill Livingstone.

Schmitz, K., Patuto, M., & Muller, N. (2012). The null-subject parameter at the interface between syntax and pragmatics: Evidence from bilingual German-Italian, German-French and Italian-French children. First Language 32(1-2), 205-238.


35

Seal, B. & Hammett, L. (1995). Language intervention with a hearing child whose parents are deaf. American Journal of Speech-Language Pathology 4(4), 15-21.

Serratrice, L. and Sorace, A. (2003). Overt and Null Subjects in Monolingual and Bilingual Italian Acquisition. Proceedings of the 27th Annual BUCLD, 739-750. Somerville, MA: Cascadilla Press.

Serratrice, L., Sorace, A., & Paoli, S. (2004). Crosslinguistic influence at the syntax-pragmatics interface: Subjects and objects in English-Italian bilingual and monolingual acquisition. Bilingualism: Language and Cognition 7 (3), 183-205. Serratrice, L., Sorace, A., Filiaci, F., & Baldo, M. (2011). Pronominal objects in English–Italian and Spanish–Italian bilingual children. Applied Psycholinguistics, 1–27. doi:10.1017/S0142716411000543

Sigurðsson, H. & Maling, J. (2010). The empty left edge condition. In M. Putnam (Ed.) Exploring Crash-Proof Grammars, 59–86. Amsterdam: John Benjamins.

Sigurðsson, H. (2011). Conditions on argument drop. Linguistic Inquiry 42, 267-394. Snyder, W., A. Senghas & K. Inman. (2001) Agreement morphology and the acquisition of

noun-drop in Spanish. Language Acquisition 9,157-173. Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic

Approaches to Bilingualism 1(1), 1–33. Suppes, P. (1974). The semantics of children’s language. American Psychologist, 29(2):103. Todd, P. (1971). A case of structural interference across sensory modalities in second-language

learning. Word 27(1-3), 102-118. Todd, P. (2008). Does ASL Really Have Just Two Grammatical Persons? Sign Language

Studies, 9(2), 166–210. Tsiimpli, I. (2011). External interfaces and the notion of ‘default.’ Linguistic Approaches to

Bilingualism 1(1), 101-103. Valian, V. (1991). Syntactic subjects in the early speech of American and Italian children.

Cognition 40, 21-81. van den Bogaerde, B., & Baker, A. (2005). Code mixing in mother-child interaction in deaf

families. Sign Language & Linguistics 8(1-2), 153-176. Wang, Q., Lillo-Martin, D., Best, C., & Levitt, A. (1992). Null subject versus null object: Some

evidence from the acquisition of Chinese and English. Language Acquisition 2, 221-254. Weir, A. (2009). Subject drop in informal English. Ms. Available online at www.isle-

linguistics.org/resources/weir2009.pdf. Accessed on April 20, 2010. Wulf, A., Dudis, P., Bayley, R., & Lucas, C. (2002). Variable Subject Presence in ASL

Narratives. Sign Language Studies 3(1), 54–76. Zwanziger, E. E., Allen, S. E. M., & Genesee, F. (2005). Cross-linguistic influence in bilingual

acquisition: Subject omission in learners of Inuktitut and English. Journal of Child Language 32(4), 893-909.

Language interaction effects in bimodal bilingualism: Argument omission in the languages of hearing ASL-English bilinguals

Documents