February 9
-
Demet Kayabaşı
-
Noyan Dokuyan & Pavel Logačev
-
Batuhan Karataş
-
Önder Güzel
<
>
Null Subjects in TID: Pro-Drop or Topic-Drop?
Demet Kayabaşı
Boğaziçi University
Phonologically null nominal arguments have been observed both in languages which employ a person agreement system and in those which do not mark person agreement.This phenomenon is analyzed as pro-drop in languages which are argued to license their null arguments via syntactic agreement, whereas it is usually considered topic-drop in languages that lack such morphology, since such null elements in those languages are only allowed when once they have been overtly introduced to the discourse as the topic of the sentence (Huang, 1984; Rapaso; 1984; Kornfilt, 1984; Enç, 1986; Rizzi 1986, Hasegawa, 1986 among others). The phenomenon of null arguments has also been observed in various sign languages such as American Sign Language, Australian Sign Language and New Zealand Sign Language (Lillo-Martin, 1986, 1991; Sandler & Lillo-Martin, 2005; Bahan, Kegl, Lee, MacLaughlin & Neidle, 2000; McKee, Schembry, Mckee & Johnstone, [2009]2012). This study aims to find out whether such phenomenon is present in Turkish Sign Language (henceforth, TID) and if that is the case, what kind of a distribution it displays. Sevinç (2006), argues that null subjects, which she considers to be pro-drop, are attested in TID too, however she argues that agentive subjects of single agreement verbs cannot be dropped and null arguments are incompatible with plain transitive verbs among Padden’S (1988 [1983]) three verb categories in sign languages, namely plain verbs, agreement verbs and spatial verbs. This study aims to show that null subjects in TID are not sensitive to the verb type they appear with, as can be seen from our data in (1). Moreover, preliminary observations indicate that null subjects in TID may in fact be instances of topic-drop instead of pro-drop. The data for this study is obtained from four video recordings of natural conversations recorded on 18.04.2013 and 30.04.2014 between two participants and story-telling tasks performed by a single signer recorded on 13.09.2012 and 25.06.2013, taken from the TIDBIL corpus.
Keywords: null arguments, pro-drop, topic-drop, Turkish Sign Language, sign language
Demet Kayabaşı
Boğaziçi University
Phonologically null nominal arguments have been observed both in languages which employ a person agreement system and in those which do not mark person agreement.This phenomenon is analyzed as pro-drop in languages which are argued to license their null arguments via syntactic agreement, whereas it is usually considered topic-drop in languages that lack such morphology, since such null elements in those languages are only allowed when once they have been overtly introduced to the discourse as the topic of the sentence (Huang, 1984; Rapaso; 1984; Kornfilt, 1984; Enç, 1986; Rizzi 1986, Hasegawa, 1986 among others). The phenomenon of null arguments has also been observed in various sign languages such as American Sign Language, Australian Sign Language and New Zealand Sign Language (Lillo-Martin, 1986, 1991; Sandler & Lillo-Martin, 2005; Bahan, Kegl, Lee, MacLaughlin & Neidle, 2000; McKee, Schembry, Mckee & Johnstone, [2009]2012). This study aims to find out whether such phenomenon is present in Turkish Sign Language (henceforth, TID) and if that is the case, what kind of a distribution it displays. Sevinç (2006), argues that null subjects, which she considers to be pro-drop, are attested in TID too, however she argues that agentive subjects of single agreement verbs cannot be dropped and null arguments are incompatible with plain transitive verbs among Padden’S (1988 [1983]) three verb categories in sign languages, namely plain verbs, agreement verbs and spatial verbs. This study aims to show that null subjects in TID are not sensitive to the verb type they appear with, as can be seen from our data in (1). Moreover, preliminary observations indicate that null subjects in TID may in fact be instances of topic-drop instead of pro-drop. The data for this study is obtained from four video recordings of natural conversations recorded on 18.04.2013 and 30.04.2014 between two participants and story-telling tasks performed by a single signer recorded on 13.09.2012 and 25.06.2013, taken from the TIDBIL corpus.
Keywords: null arguments, pro-drop, topic-drop, Turkish Sign Language, sign language
Missing-Verb Effects in a Head-Final Language
Noyan Dokudan, Pavel Logačev
Boğaziçi University, Istanbul,
Nested structures usually cause great processing difficulty. Surprisingly, omitting one verb from such complex structures, as in (1), can sometimes ease processing, although doing so renders the sentence ungrammatical. This so-called missing-VP effect surfaces during faster reading of sentences with a missing verb (Gimenes et al., 2009; Vasisth et al. 2010; Frank et al., 2016). Such sentences are also perceived as equally or less complex (Gibson & Thomas, 1999; Gimenes et al., 2009), and are sometimes considered acceptable. (Hӓussler & Bader, 2015)
Häussler & Bader (2015) argue that an illusion of grammaticality occurs as a result of interference during cue-based retrieval. We propose an alternative account based on interference and theta-role assignment: In missing-verb versions of (1), NP2 (student) is assigned a theta-role by confused, which sometimes results in the erroneous retrieval of NP1 (manuscript) during the processing of the last verb. The resulting parse can cause an illusion of grammaticality, due to the fact that each NP has at least one theta-role. Our hypothesis predicts that sentences with missing verbs are more likely to be judged as acceptable when the subject of that verb has received a theta-role from another source, such as a relative clause (RC) modifying it. To test this hypothesis, we conducted a speeded acceptability judgment experiment in Turkish (N=78) with sentences containing center-embedded complement clauses. All 48 experimental items had the structure of the example in (2). Ungrammatical missing-verb sentences were derived from their grammatical counterparts by omitting the second embedded verb (V2; listen to). All subject nouns were either (i) modified by an RC (theta-assignment conditions), or (ii) modified by an adjective (control conditions with modification), or (iii) bare (simple control conditions).52 filler sentences were designed to prevent noun- or verb-counting strategies.
‘I know that the (young) journalist (who visited recently) filmed the (old) cop (who died today) listening to the (agitated) teacher (who got ill) speak.
The percentages of ‘acceptable’ responses for grammatical and ungrammatical sentences respectively were 81% and 65% in bare conditions, 81% and 76% in adjective conditions, 83% and 73% in RC conditions, and 91% and 8% for fillers. For experimental items, a generalized linear mixed-effects model showed an effect of grammaticality (β=.83, SE=.15, z=5.4), and a significant interaction between grammaticality and the presence of modification (β=.4, SE=.13, z=2.9), but no interaction between grammaticality and modification type (β=-.4, SE=0.26, z=-1.5). The approximately equal error rates for both grammatical and ungrammatical filler sentences suggest that the high percentage of ‘acceptable’ responses in ungrammatical conditions might be due to a grammaticality illusion instead of a response bias. Thus, our results are the first demonstration of a missing-verb effect in a strictly head-final language.
Noyan Dokudan, Pavel Logačev
Boğaziçi University, Istanbul,
Nested structures usually cause great processing difficulty. Surprisingly, omitting one verb from such complex structures, as in (1), can sometimes ease processing, although doing so renders the sentence ungrammatical. This so-called missing-VP effect surfaces during faster reading of sentences with a missing verb (Gimenes et al., 2009; Vasisth et al. 2010; Frank et al., 2016). Such sentences are also perceived as equally or less complex (Gibson & Thomas, 1999; Gimenes et al., 2009), and are sometimes considered acceptable. (Hӓussler & Bader, 2015)
- [S1The manuscript [S2that the student [S3who the catalog confused] was studying] was missing a page].
Häussler & Bader (2015) argue that an illusion of grammaticality occurs as a result of interference during cue-based retrieval. We propose an alternative account based on interference and theta-role assignment: In missing-verb versions of (1), NP2 (student) is assigned a theta-role by confused, which sometimes results in the erroneous retrieval of NP1 (manuscript) during the processing of the last verb. The resulting parse can cause an illusion of grammaticality, due to the fact that each NP has at least one theta-role. Our hypothesis predicts that sentences with missing verbs are more likely to be judged as acceptable when the subject of that verb has received a theta-role from another source, such as a relative clause (RC) modifying it. To test this hypothesis, we conducted a speeded acceptability judgment experiment in Turkish (N=78) with sentences containing center-embedded complement clauses. All 48 experimental items had the structure of the example in (2). Ungrammatical missing-verb sentences were derived from their grammatical counterparts by omitting the second embedded verb (V2; listen to). All subject nouns were either (i) modified by an RC (theta-assignment conditions), or (ii) modified by an adjective (control conditions with modification), or (iii) bare (simple control conditions).52 filler sentences were designed to prevent noun- or verb-counting strategies.
- Ben biliyorum [S1 [(geçen gelen / genç) haberci-nin] [S2 [(bugün öl-en / yaşlı) polisin] I know recently come-REL / young journalist-GEN today die-REL / old cop [S3 [(hastalanan / tedirgin) öğretmen-in] konuşmasını] dinlemesini] görüntülediğini.
‘I know that the (young) journalist (who visited recently) filmed the (old) cop (who died today) listening to the (agitated) teacher (who got ill) speak.
The percentages of ‘acceptable’ responses for grammatical and ungrammatical sentences respectively were 81% and 65% in bare conditions, 81% and 76% in adjective conditions, 83% and 73% in RC conditions, and 91% and 8% for fillers. For experimental items, a generalized linear mixed-effects model showed an effect of grammaticality (β=.83, SE=.15, z=5.4), and a significant interaction between grammaticality and the presence of modification (β=.4, SE=.13, z=2.9), but no interaction between grammaticality and modification type (β=-.4, SE=0.26, z=-1.5). The approximately equal error rates for both grammatical and ungrammatical filler sentences suggest that the high percentage of ‘acceptable’ responses in ungrammatical conditions might be due to a grammaticality illusion instead of a response bias. Thus, our results are the first demonstration of a missing-verb effect in a strictly head-final language.
“falan” Dilsel Biriminin Sözcük Türü Üzerine
Batuhan Karataş
Dokuz Eylül University
Belirli bir sözcük türünde tanımlanmış her dilsel birim, o sözcük türünün biçimbilimsel ve sözdizimsel özelliklerine uygun davranmalıdır. Alanyazında ˈfalanˈ dilsel biriminin sözcük türüne yönelik farklı saptamalar yer almaktadır. Sözgelimi, bu sözcüğün Kornfilt (1997) tarafından ünlem olduğu, Lewis (2000) tarafından adıl ve sıfat olduğu, Dil Derneği (2017) tarafından ad, sıfat ve adıl olduğu savlanmıştır. ˈfalanˈ dilsel birimi üzerine alanyazındaki bu uzlaşmazlık, hem bilimsel çalışmalar hem de eğitim çalışmaları için olası bir sorundur. Çalışma, alanyazındaki bu görüş farklılığına açıklık getirmeyi, inceleme nesnesinin sözcük türünü ortaya koymayı amaçlamaktadır. Bu amaç doğrultusunda öncelikle bu sözcük bağlam-duyarsız tümcelerle sınanacak ve sözcük ulamına yönelik bir öngörü elde edilecektir. Sözgelimi, ˈfalanˈ dilsel birimi Dil Derneği (2017) tarafından savlandığı gibi bir ad ise beklenen, bu dilsel birimin örneğin durum biçimbirimi ile birlikte kullanılabilmesidir:
(2) Ayşei[Alik falan-ı*i/*kgördü] sanmış.
Bu birimin ne iç tümce öznesini ne de dış tümce öznesini bağlayabilmesi (Chomsky, 1986), adılsıl ya da gönderge olmadığını göstermektedir. Elde edilen bu gibi gözlemlerin geçerliliğini sınamak ve olası yeni bulgulara ulaşmak için yukarıdaki gibi bağlamdan soyut bir sınamanın yanı sıra derlem tabanlı bir inceleme yapılacaktır:
(3a ve b)’deki gibi veriler ise falan dilsel biriminin ilgeç ulamında da yer alması gerektiğini göstermektedir. Dolayısıyla “falan” dilsel biriminin sözcük türünü belirlemeye yönelik yapılan bu çalışmanın iki boyutunun olduğu söylenebilir: (i) biçimbilimsel ve sözdizimsel sınamalar ve (ii) edim temelli incelenme. Bu öngözlemler doğrultusunda, ˈfalanˈ dilsel biriminin öncelikle bağlam-duyarsız sınamalarla hangi dilsel ulam olamayacağı; sonrasındaki derlem-temelli inceleme ile hangi ulama ait olabileceği belirlenecektir.
Anahtar Sözcükler: falan, sözcük türü, dağılım ve yerine koyma teknikleri, derlem-temelli inceleme.
Batuhan Karataş
Dokuz Eylül University
Belirli bir sözcük türünde tanımlanmış her dilsel birim, o sözcük türünün biçimbilimsel ve sözdizimsel özelliklerine uygun davranmalıdır. Alanyazında ˈfalanˈ dilsel biriminin sözcük türüne yönelik farklı saptamalar yer almaktadır. Sözgelimi, bu sözcüğün Kornfilt (1997) tarafından ünlem olduğu, Lewis (2000) tarafından adıl ve sıfat olduğu, Dil Derneği (2017) tarafından ad, sıfat ve adıl olduğu savlanmıştır. ˈfalanˈ dilsel birimi üzerine alanyazındaki bu uzlaşmazlık, hem bilimsel çalışmalar hem de eğitim çalışmaları için olası bir sorundur. Çalışma, alanyazındaki bu görüş farklılığına açıklık getirmeyi, inceleme nesnesinin sözcük türünü ortaya koymayı amaçlamaktadır. Bu amaç doğrultusunda öncelikle bu sözcük bağlam-duyarsız tümcelerle sınanacak ve sözcük ulamına yönelik bir öngörü elde edilecektir. Sözgelimi, ˈfalanˈ dilsel birimi Dil Derneği (2017) tarafından savlandığı gibi bir ad ise beklenen, bu dilsel birimin örneğin durum biçimbirimi ile birlikte kullanılabilmesidir:
- a. *Ali falan-ı odaya getirdi. b. *Ali falan-dan düştü.
(2) Ayşei[Alik falan-ı*i/*kgördü] sanmış.
Bu birimin ne iç tümce öznesini ne de dış tümce öznesini bağlayabilmesi (Chomsky, 1986), adılsıl ya da gönderge olmadığını göstermektedir. Elde edilen bu gibi gözlemlerin geçerliliğini sınamak ve olası yeni bulgulara ulaşmak için yukarıdaki gibi bağlamdan soyut bir sınamanın yanı sıra derlem tabanlı bir inceleme yapılacaktır:
- a. “Ee anlat bakalım, kurbağa, yılan falan yediniz mi hiç? – Yok” b. “ – Çatışma falan yaşandı mı? Hüseyin: Çok mecbur”
(3a ve b)’deki gibi veriler ise falan dilsel biriminin ilgeç ulamında da yer alması gerektiğini göstermektedir. Dolayısıyla “falan” dilsel biriminin sözcük türünü belirlemeye yönelik yapılan bu çalışmanın iki boyutunun olduğu söylenebilir: (i) biçimbilimsel ve sözdizimsel sınamalar ve (ii) edim temelli incelenme. Bu öngözlemler doğrultusunda, ˈfalanˈ dilsel biriminin öncelikle bağlam-duyarsız sınamalarla hangi dilsel ulam olamayacağı; sonrasındaki derlem-temelli inceleme ile hangi ulama ait olabileceği belirlenecektir.
Anahtar Sözcükler: falan, sözcük türü, dağılım ve yerine koyma teknikleri, derlem-temelli inceleme.
{-(s)Al} mı? {-(A)l} mı?
Önder Güzel
Dokuz Eylül University
Çağdaş biçimbilim incelemelerinde eklendiği dilsel birime “ilgili, olan, ile, benzeri” gibi anlamlar kattığı savlanan -(s)Al ya da -(A)l biçimbiriminin kökenine ilişkin bir görüş birliği bulunmadığı gibi, söz konusu biçimbirimlerin üstyazımında da farklı yönelimler bulunmaktadır. Örneğin, Göksel ve Kerslake (2005) -sAl biçiminde bir üstyazımı savunmakta ve -(A)l biçimbiriminin -sAl biçimbirimiyle aynı işleve sahip olduğunu savlamaktadır. Kornfilt (1997) söz konusu biçimbirim için -(s)Al, Lewis (2000) ise -(s)el üstyazımını kullanmaktadır. Buna ek olarak Lewis’in ve ilgili konudaki görüşünün takipçisi Kornfilt de, son sesbirimi /s/ veya /z/ olan sözcüklere, söz konusu biçimbirimde yer alan /s/ sesbiriminin eklenmediği görüşünde hemfikirdirler; ancak Türkçede araştırmacıların bu görüşüne koşut özel gibi sözcüklerin yanı sıra, bu görüşe uymayan ussal, sessel, özsel gibi sözcükler de bulunmaktadır. Bu çalışmanın amacı, yukarıdaki tartışmalar çerçevesinde sözü edilen biçimbirimin -(s)Al mı, -(A)l mı olduğunu belirlemektir. Bu amaçla Belge Tarama (Document Analysis) veri toplama yöntemiyle Dil Derneği’ne (2012) ait olan Türkçe Sözlük kullanılmış, burada sonu /s/ veya /z/ sesbirimiyle biten ve üzerine -(s)Al alan tüm sözcükler dizelgelenmiştir. Daha sonra dizelgelenen sözcüklerden söz konusu biçimbirimin sözcük türüne, eklendiği sözcüğün anlambilimsel özelliklerine ve sesbilimsel etmenlere göre sona eklenirken biçimlenişi betimlenmiştir. Bunun sonucunda da yapılan betiklerden elde edilen bulgularla hangi üstyazımın yeğlenmesinin daha akılcıl sonuçlar sunacağı değerlendirilecektir ve söz konusu biçimbirimin üstyazımı konusunda daha akılcıl sonuçların tespit edilmesi sözcük türetmede işe yarayacağı gibi söz konusu biçimbirimlerden daha işlek olanını ortaya koyma açısından da önemlidir. Sonuç olarak, inceleme konusu olan biçimbirim genelde sözcük türü ad olan tabanlara gelerek onları sıfat yapmaktadır ve dolayısıyla sonu /s/ ve /z/ sesleri ile biten sözcüklere eklendiğinde de sıfat türetimi olduğu gözlemlenmiştir. Araştırma sonucunda –(A)l ve –(s)Al biçimbirimleri aynı sözcüğe eklendiğinde ortaya çıkan başka sözcüklerin birbirinden bağımsız oldukları bu yüzden de iki adet biçimbirim olduğu saptanmıştır.
Anahtar Sözcükler: türetim biçimbirimi, sözlüksel biçimbilim, üstyazım, belge
Önder Güzel
Dokuz Eylül University
Çağdaş biçimbilim incelemelerinde eklendiği dilsel birime “ilgili, olan, ile, benzeri” gibi anlamlar kattığı savlanan -(s)Al ya da -(A)l biçimbiriminin kökenine ilişkin bir görüş birliği bulunmadığı gibi, söz konusu biçimbirimlerin üstyazımında da farklı yönelimler bulunmaktadır. Örneğin, Göksel ve Kerslake (2005) -sAl biçiminde bir üstyazımı savunmakta ve -(A)l biçimbiriminin -sAl biçimbirimiyle aynı işleve sahip olduğunu savlamaktadır. Kornfilt (1997) söz konusu biçimbirim için -(s)Al, Lewis (2000) ise -(s)el üstyazımını kullanmaktadır. Buna ek olarak Lewis’in ve ilgili konudaki görüşünün takipçisi Kornfilt de, son sesbirimi /s/ veya /z/ olan sözcüklere, söz konusu biçimbirimde yer alan /s/ sesbiriminin eklenmediği görüşünde hemfikirdirler; ancak Türkçede araştırmacıların bu görüşüne koşut özel gibi sözcüklerin yanı sıra, bu görüşe uymayan ussal, sessel, özsel gibi sözcükler de bulunmaktadır. Bu çalışmanın amacı, yukarıdaki tartışmalar çerçevesinde sözü edilen biçimbirimin -(s)Al mı, -(A)l mı olduğunu belirlemektir. Bu amaçla Belge Tarama (Document Analysis) veri toplama yöntemiyle Dil Derneği’ne (2012) ait olan Türkçe Sözlük kullanılmış, burada sonu /s/ veya /z/ sesbirimiyle biten ve üzerine -(s)Al alan tüm sözcükler dizelgelenmiştir. Daha sonra dizelgelenen sözcüklerden söz konusu biçimbirimin sözcük türüne, eklendiği sözcüğün anlambilimsel özelliklerine ve sesbilimsel etmenlere göre sona eklenirken biçimlenişi betimlenmiştir. Bunun sonucunda da yapılan betiklerden elde edilen bulgularla hangi üstyazımın yeğlenmesinin daha akılcıl sonuçlar sunacağı değerlendirilecektir ve söz konusu biçimbirimin üstyazımı konusunda daha akılcıl sonuçların tespit edilmesi sözcük türetmede işe yarayacağı gibi söz konusu biçimbirimlerden daha işlek olanını ortaya koyma açısından da önemlidir. Sonuç olarak, inceleme konusu olan biçimbirim genelde sözcük türü ad olan tabanlara gelerek onları sıfat yapmaktadır ve dolayısıyla sonu /s/ ve /z/ sesleri ile biten sözcüklere eklendiğinde de sıfat türetimi olduğu gözlemlenmiştir. Araştırma sonucunda –(A)l ve –(s)Al biçimbirimleri aynı sözcüğe eklendiğinde ortaya çıkan başka sözcüklerin birbirinden bağımsız oldukları bu yüzden de iki adet biçimbirim olduğu saptanmıştır.
Anahtar Sözcükler: türetim biçimbirimi, sözlüksel biçimbilim, üstyazım, belge
February 10
-
Berna Uzundağ & Aylin Küntay
-
Ecem Baykuş
-
Tzu-Ching Kao
-
Fidan Can, Nermin Boğaz & Nazan Bilgin
<
>
The Acquisition and Use of Relative Clauses in Turkish: An Analysis of Spontaneous Interactions
Berna A. Uzundağ & Aylin C. Küntay
Koç University
Children’s acquisition of relative clauses has been researched as a representative case of the acquisition of complex language structures. We investigated Turkish-speaking children’s developmental trajectory of relative clauses in relation to child-directed speech by using longitudinal and cross-sectional corpora. In the longitudinal corpus, interactions between 8 children and their caregivers (4 high- and 4 low-SES families) were recorded bimonthly between 8 and 36 months (Küntay et al., 2015). In the cross-sectional corpus, peer interactions within triads of 78 children (43 to 64 months of age) were recorded while they were engaging in activities such as drawing and playing with memory cards (Köymen, 2005). We coded each utterance in child and child-directed speech that contained a relative clause (N=618) according to (a) the syntactic role of the head noun in the relative clause (subject, object, oblique, and genitive relative), (b) the syntactic role of the head noun in the matrix clause (subject, object, oblique element, adjunct, isolated NP, and predicate nominal), (c) type of subject (pronominal or lexical), and (d) the animacy of the head noun. In early child speech (before 36 months), relative clauses were rare and just emerging (N=27), corroborating previous findings that Turkish relative clauses are produced late (e.g., Altınkamış & Altan, 2016). The frequency of relative clauses in caregiver input was very low as well (0.26%), but high-SES caregivers produced more relative clauses than low-SES caregivers, t(6)=4.61, p=.004. In early child speech, subject relative clauses, being morphosyntactically simpler than other types (Slobin, 1986), were the most frequent type (see Figure 1) and tended to include animate agents. Object relative clauses were the most frequent category in late child speech and caregiver input, and almost always occurred with inanimate head nouns and pronominal subjects; a pattern known to reduce their processing difficulty (e.g., Kidd et al., 2007) (see Figure 2). Although children mostly used relative clauses as the object or subject of the matrix clauses (see Figure 3), the constructions that relative clauses appeared in were simpler than the ones in child-directed speech in that they mostly either conveyed a single proposition (e.g., ormanda yaşayan bir ördek var ‘there is a duck that lives in the forest’) or had a light verb (e.g., yap- ‘do’). Overall, the results are consistent with findings in typologically different languages (e.g., Chen & Shirai, 2015), suggesting that the acquisition of relative clauses is affected by the input and is a gradual development.
Berna A. Uzundağ & Aylin C. Küntay
Koç University
Children’s acquisition of relative clauses has been researched as a representative case of the acquisition of complex language structures. We investigated Turkish-speaking children’s developmental trajectory of relative clauses in relation to child-directed speech by using longitudinal and cross-sectional corpora. In the longitudinal corpus, interactions between 8 children and their caregivers (4 high- and 4 low-SES families) were recorded bimonthly between 8 and 36 months (Küntay et al., 2015). In the cross-sectional corpus, peer interactions within triads of 78 children (43 to 64 months of age) were recorded while they were engaging in activities such as drawing and playing with memory cards (Köymen, 2005). We coded each utterance in child and child-directed speech that contained a relative clause (N=618) according to (a) the syntactic role of the head noun in the relative clause (subject, object, oblique, and genitive relative), (b) the syntactic role of the head noun in the matrix clause (subject, object, oblique element, adjunct, isolated NP, and predicate nominal), (c) type of subject (pronominal or lexical), and (d) the animacy of the head noun. In early child speech (before 36 months), relative clauses were rare and just emerging (N=27), corroborating previous findings that Turkish relative clauses are produced late (e.g., Altınkamış & Altan, 2016). The frequency of relative clauses in caregiver input was very low as well (0.26%), but high-SES caregivers produced more relative clauses than low-SES caregivers, t(6)=4.61, p=.004. In early child speech, subject relative clauses, being morphosyntactically simpler than other types (Slobin, 1986), were the most frequent type (see Figure 1) and tended to include animate agents. Object relative clauses were the most frequent category in late child speech and caregiver input, and almost always occurred with inanimate head nouns and pronominal subjects; a pattern known to reduce their processing difficulty (e.g., Kidd et al., 2007) (see Figure 2). Although children mostly used relative clauses as the object or subject of the matrix clauses (see Figure 3), the constructions that relative clauses appeared in were simpler than the ones in child-directed speech in that they mostly either conveyed a single proposition (e.g., ormanda yaşayan bir ördek var ‘there is a duck that lives in the forest’) or had a light verb (e.g., yap- ‘do’). Overall, the results are consistent with findings in typologically different languages (e.g., Chen & Shirai, 2015), suggesting that the acquisition of relative clauses is affected by the input and is a gradual development.
Any Loss in Pro-Drop Feature? Null and Overt Subject Usages of Turkish Heritage Speakers
Ecem Baykuş
METU
Heritage speakers are bilinguals who are exposed to both the heritage language (it can also be called ethnic, minority, or family language) at home early in childhood and to the majority language intensively which is the language of the society they live in and the education that they get (Long, M. H., & Doughty, C. J. (Eds.), 2011). It is argued that due to the dual language environment and the limited exposure to the heritage language, there is an incomplete acquisition of the heritage language (Montrul, 2009). This study examines Turkish heritage speakers’ comprehension and production of the pro-drop feature in Turkish to investigate whether there is any incomplete acquisition of Turkish pro-drop feature in Turkish heritage speakers. Although heritage speakers’ phonology, lexicon, and morphology are affected by the limited exposure (Au et all, 2002; Polinsky, 2005; Montrul, 2004), Montrul hypothesizes that” syntax is impervious to language loss” (Montrul, 2004). We tested the applicability of Montrul’s hypothesis on Turkish heritage speakers raised and living in Germany. They are exposed to German as the dominant language while Turkish remains as their minority language The research instruments consist of two tasks; naturality judgment task (NJT) and a picture description task (PDT). The purpose of NJT was to examine their perception and competence on overt and null subjects in Turkish. NJT consisted of grammatical Turkish sentences such as ‘Sen uyurken bu elmaların hepsini ağaçtan ben topladım.‘ as well as the ungrammatical ones like ‘Dondurma isteyen var mı aranızda ?*- İsterim.’. The purpose of PDT was to examine their written production of null and overt subjects in a context. For the scope of this research, two experiments were conducted. In the first experiment, monolingual Turkish speakers completed the tasks, which were also administered to the heritage speakers in the second experiment, so that monolingual speakers linguistic command on pro-drop feature is identified. In the second experiment, Turkish heritage speakers were given the same tasks to analyze their comprehension and production on Turkish pro-drop feature. Later, the findings of these two experiments were compared, and it revealed that Turkish heritage speakers showed native-like linguistic behavior in both the comprehension and production tasks. The results also imply that the heritage speakers reached complete acquisition in the pro-drop feature in their heritage language. In other words, Turkish heritage speakers maintain Turkish pro-drop feature although they are dominantly exposed to a non-pro-drop language, German. Overall this study supports Montruls’ hypothesis that syntax is not vulnerable to language loss among heritage speakers.
Key words: heritage language acquisition, Turkish heritage speakers, pro-drop feature, syntax
Ecem Baykuş
METU
Heritage speakers are bilinguals who are exposed to both the heritage language (it can also be called ethnic, minority, or family language) at home early in childhood and to the majority language intensively which is the language of the society they live in and the education that they get (Long, M. H., & Doughty, C. J. (Eds.), 2011). It is argued that due to the dual language environment and the limited exposure to the heritage language, there is an incomplete acquisition of the heritage language (Montrul, 2009). This study examines Turkish heritage speakers’ comprehension and production of the pro-drop feature in Turkish to investigate whether there is any incomplete acquisition of Turkish pro-drop feature in Turkish heritage speakers. Although heritage speakers’ phonology, lexicon, and morphology are affected by the limited exposure (Au et all, 2002; Polinsky, 2005; Montrul, 2004), Montrul hypothesizes that” syntax is impervious to language loss” (Montrul, 2004). We tested the applicability of Montrul’s hypothesis on Turkish heritage speakers raised and living in Germany. They are exposed to German as the dominant language while Turkish remains as their minority language The research instruments consist of two tasks; naturality judgment task (NJT) and a picture description task (PDT). The purpose of NJT was to examine their perception and competence on overt and null subjects in Turkish. NJT consisted of grammatical Turkish sentences such as ‘Sen uyurken bu elmaların hepsini ağaçtan ben topladım.‘ as well as the ungrammatical ones like ‘Dondurma isteyen var mı aranızda ?*- İsterim.’. The purpose of PDT was to examine their written production of null and overt subjects in a context. For the scope of this research, two experiments were conducted. In the first experiment, monolingual Turkish speakers completed the tasks, which were also administered to the heritage speakers in the second experiment, so that monolingual speakers linguistic command on pro-drop feature is identified. In the second experiment, Turkish heritage speakers were given the same tasks to analyze their comprehension and production on Turkish pro-drop feature. Later, the findings of these two experiments were compared, and it revealed that Turkish heritage speakers showed native-like linguistic behavior in both the comprehension and production tasks. The results also imply that the heritage speakers reached complete acquisition in the pro-drop feature in their heritage language. In other words, Turkish heritage speakers maintain Turkish pro-drop feature although they are dominantly exposed to a non-pro-drop language, German. Overall this study supports Montruls’ hypothesis that syntax is not vulnerable to language loss among heritage speakers.
Key words: heritage language acquisition, Turkish heritage speakers, pro-drop feature, syntax
The role of lexical categories in reading Chinese
Tzu-Ching Kao
Middle East Technical University
This study aims to observe the language processing behavior of Chinese separable words (SWs) based on Combinatory Categorial Grammar (CCG). The study is currently at early stage that we focus on deep analyses on semantic and syntactic interaction of SWs to form lexical categories for later eye-tracking experiments. There are at least two advantages of CCG to tackle SWs. First, the so-called interface among semantics, syntax, and intonation, etc. are all projected into lexical categories, which enable us to transform their interaction of lexical items in a transparent and assessable way. Incremental processing as one of the hypotheses in CCG, each constituent has to deliver the syntactic and semantic result according to the lexical categories and universal combination rules. It establishes a research niche to test eye movements in reading as they are also processed in an incremental way. SWs words are notoriously known about their dual status of wordhood. They are conceptually perceived as one single lexical entry but can be split through syntactic operations, e.g. taking aspect markers, topicalization, etc. Until very recently, studies about SWs have been primarily concerned with the wordhood from a pure syntactic perspective. In this paper, we argue that the dichotomic method may overlook their underlying semantic di erences between split and integral forms. The internal verb-object structure and event structure of SWs appear much more relevant to tackle the apparatus of their heterogeneous separability. Following this route, I examined the distinction of split and integral forms of SWs with the typology of event con ation theory(Talmy, 1999) and event structure and argue that their syntactic di erence is a result of semantic level.
Keywords: separable words in Chinese, semantics, syntax, Combinatory Categorial Grammar
Tzu-Ching Kao
Middle East Technical University
This study aims to observe the language processing behavior of Chinese separable words (SWs) based on Combinatory Categorial Grammar (CCG). The study is currently at early stage that we focus on deep analyses on semantic and syntactic interaction of SWs to form lexical categories for later eye-tracking experiments. There are at least two advantages of CCG to tackle SWs. First, the so-called interface among semantics, syntax, and intonation, etc. are all projected into lexical categories, which enable us to transform their interaction of lexical items in a transparent and assessable way. Incremental processing as one of the hypotheses in CCG, each constituent has to deliver the syntactic and semantic result according to the lexical categories and universal combination rules. It establishes a research niche to test eye movements in reading as they are also processed in an incremental way. SWs words are notoriously known about their dual status of wordhood. They are conceptually perceived as one single lexical entry but can be split through syntactic operations, e.g. taking aspect markers, topicalization, etc. Until very recently, studies about SWs have been primarily concerned with the wordhood from a pure syntactic perspective. In this paper, we argue that the dichotomic method may overlook their underlying semantic di erences between split and integral forms. The internal verb-object structure and event structure of SWs appear much more relevant to tackle the apparatus of their heterogeneous separability. Following this route, I examined the distinction of split and integral forms of SWs with the typology of event con ation theory(Talmy, 1999) and event structure and argue that their syntactic di erence is a result of semantic level.
Keywords: separable words in Chinese, semantics, syntax, Combinatory Categorial Grammar
An Investigation Into the Effect of Orthographic Sensitivity on The Conjugation of Past Tense Morphemes In L2
Fidan Can, Nazan Bilgin, Nermin Boğaz
Middle East Technical University
The study aims to investigate the effect of orthographic sensitivity, “sensitivity to regularities of orthographic structure” of written items (Vellutino, Scanlon & Tanzman, 1994, p.251 as cited in Andrews, 2008) in L2 while conjugating verbs in Simple Past Tense. While regular verbs are rule-governed, irregular verbs require memory retrieval (Pinker, 1999). Ullman (2006) supports this argument in a different way by using different concepts; procedural memory and declarative memory. He proposes that declarative memory is the only long-term memory which is responsible for explicit knowledge, and therefore we must have stored the features of lexical entries in declarative memory, while by means of procedural memory, only the tacit sequences and rules are supposed to be kept. Regarding this view, the declarative memory is the place where the information about irregular past tense forms is stored (Ullman, 2006). However, it is stated that procedural memory is less available to the learners of L2. Their implicit competence includes fewer items compared to that of native speakers of English (Paradis, 2009 as cited in Faruji, 2012). As L2 learners gain exposure, experience, and proficiency with the L2, they get to depend on procedural memory (Squire & Zola, 1996 as cited in Morgan-Short & Briel-Schuetz, 2014). In addition, it is also stated that the success in advanced phases of L2 development can rely especially on procedural memory (Ullman, DeKeyser, Paradis, 2005). Because irregular verb forms are stored in declarative memory, when the learners come across pseudo verbs whose spellings are similar to real irregular verbs, it can be assumed that they are prone to retrieve knowledge of conjugating these pseudo irregular verbs from this declarative memory because of the orthographic similarity (e.g.: bleed - sleed). We utilized quantitative and qualitative methods. For quantitative data collection, we used a language test where there were some pseudo-verbs hidden among the real verbs. For qualitative data collection, we used the introspective method; think-aloud protocol where the participants were asked to explain why they conjugated the verbs in the way they did. Elementary and advanced learners at METU participated in our research. The result is that the advanced learners utilized -ed form more than the elementary learners, which shows that the advanced learners relied on procedural memory while conjugating pseudo verbs because their explicit knowledge may have turned to implicit knowledge (proceduralization). However, we may not talk about the full access to procedural memory since they were probably under the influence of their explicit knowledge, irregular verb forms while conjugating some of the verbs [treak-treaked (troke)].
Keywords: L2, ortographic sensitivity, past tens, morphology, verb conjugation
Fidan Can, Nazan Bilgin, Nermin Boğaz
Middle East Technical University
The study aims to investigate the effect of orthographic sensitivity, “sensitivity to regularities of orthographic structure” of written items (Vellutino, Scanlon & Tanzman, 1994, p.251 as cited in Andrews, 2008) in L2 while conjugating verbs in Simple Past Tense. While regular verbs are rule-governed, irregular verbs require memory retrieval (Pinker, 1999). Ullman (2006) supports this argument in a different way by using different concepts; procedural memory and declarative memory. He proposes that declarative memory is the only long-term memory which is responsible for explicit knowledge, and therefore we must have stored the features of lexical entries in declarative memory, while by means of procedural memory, only the tacit sequences and rules are supposed to be kept. Regarding this view, the declarative memory is the place where the information about irregular past tense forms is stored (Ullman, 2006). However, it is stated that procedural memory is less available to the learners of L2. Their implicit competence includes fewer items compared to that of native speakers of English (Paradis, 2009 as cited in Faruji, 2012). As L2 learners gain exposure, experience, and proficiency with the L2, they get to depend on procedural memory (Squire & Zola, 1996 as cited in Morgan-Short & Briel-Schuetz, 2014). In addition, it is also stated that the success in advanced phases of L2 development can rely especially on procedural memory (Ullman, DeKeyser, Paradis, 2005). Because irregular verb forms are stored in declarative memory, when the learners come across pseudo verbs whose spellings are similar to real irregular verbs, it can be assumed that they are prone to retrieve knowledge of conjugating these pseudo irregular verbs from this declarative memory because of the orthographic similarity (e.g.: bleed - sleed). We utilized quantitative and qualitative methods. For quantitative data collection, we used a language test where there were some pseudo-verbs hidden among the real verbs. For qualitative data collection, we used the introspective method; think-aloud protocol where the participants were asked to explain why they conjugated the verbs in the way they did. Elementary and advanced learners at METU participated in our research. The result is that the advanced learners utilized -ed form more than the elementary learners, which shows that the advanced learners relied on procedural memory while conjugating pseudo verbs because their explicit knowledge may have turned to implicit knowledge (proceduralization). However, we may not talk about the full access to procedural memory since they were probably under the influence of their explicit knowledge, irregular verb forms while conjugating some of the verbs [treak-treaked (troke)].
Keywords: L2, ortographic sensitivity, past tens, morphology, verb conjugation
-
Simge Topaloğlu
-
Münir Özturhan
-
Pınar Kolancalı & Sonali Nag
<
>
Acquisition of {-dA}: an ambiguous particle that signals both additive focus and contrastive topic
Simge Topaloğlu
Boğaziçi University
Crosslinguistic research on preschoolers' comprehension of the existential presuppositions invoked by additive focus-particles yielded mixed results (Hüttner et al. 2004, Matsuoka 2004, Bergsma 2006, Höhle et al. 2009, Berger&Höhle 2012). Turkish presents additional challenges to children, since it has a particle, {-dA}, functioning both as an additive focus-particle and a topic-change-marker - which are disambiguated prosodically(Göksel&Özsoy 2003). When {-dA} is an additive focus-particle signalling subject-focus, the subject- NP bears stress(1a). When it is a topic-change-marker designating the subject-NP as the contrastive topic, the subject-NP is not stressed, there is a pause after the topic-change-marker, and the object-NP, (i.e.,the focus), bears stress(1b). Though both (1a&b) have the same assertion (i.e.,''Kermit is drinking tea''), they have different presuppositions; (1a) presupposes that the predicate is also true of an individual other than Kermit, (1b) presupposes that it is false of the other individual.
We investigate whether Turkish-speaking children utilize prosody to process the presuppositions of these two {-dA}'s. The experimental task requires that children choose the pictures that show what two characters are doing, based on the test sentence(Table 2). The results show that G1 and G2 tend to misinterpret the additive {-dA} as a topic-change-marker, suggesting that young children's object-focus-bias detected in other contexts (Müller 2010, Kim 2011), and which we replicated in a previous study in Turkish, may also be present here, leading preschoolers to misinterpret subject-focus sentences as subject-topic sentences. In G3, topic- change-marker overgeneralizations decline and children seem to have become proficient with the additive focus-particle. Instead, these children misinterpret over ¼ of all topic-change-markers as additives; though G4 could distinguish between both types of {-dA}. Given children's sensitivity to the distributional regularities in the input, if additive {-dA} is more frequent in child-directed speech than topic-change-marking {-dA}, older children, having outgrown their object-focus-bias, may overgeneralize the additive reading. We transcribed 3 episodes of a children's TV-show intended for the age-groups we tested. There were 164 {-dA}'s in this mini- corpus, 77 of which (47%) were additive focus-particles, while only 43 (26%) were topic-change-markers and 44 (27%) were not relevant for our test.1 As additive {-dA}'s are predominant in the input, children might misperceive additivity as the primary function of {-dA} and misanalyze some topic-change-markers as additives.
Keywords: Acquisition of Turkish, language development, additive focus, contrastive topic, prosody
Simge Topaloğlu
Boğaziçi University
Crosslinguistic research on preschoolers' comprehension of the existential presuppositions invoked by additive focus-particles yielded mixed results (Hüttner et al. 2004, Matsuoka 2004, Bergsma 2006, Höhle et al. 2009, Berger&Höhle 2012). Turkish presents additional challenges to children, since it has a particle, {-dA}, functioning both as an additive focus-particle and a topic-change-marker - which are disambiguated prosodically(Göksel&Özsoy 2003). When {-dA} is an additive focus-particle signalling subject-focus, the subject- NP bears stress(1a). When it is a topic-change-marker designating the subject-NP as the contrastive topic, the subject-NP is not stressed, there is a pause after the topic-change-marker, and the object-NP, (i.e.,the focus), bears stress(1b). Though both (1a&b) have the same assertion (i.e.,''Kermit is drinking tea''), they have different presuppositions; (1a) presupposes that the predicate is also true of an individual other than Kermit, (1b) presupposes that it is false of the other individual.
We investigate whether Turkish-speaking children utilize prosody to process the presuppositions of these two {-dA}'s. The experimental task requires that children choose the pictures that show what two characters are doing, based on the test sentence(Table 2). The results show that G1 and G2 tend to misinterpret the additive {-dA} as a topic-change-marker, suggesting that young children's object-focus-bias detected in other contexts (Müller 2010, Kim 2011), and which we replicated in a previous study in Turkish, may also be present here, leading preschoolers to misinterpret subject-focus sentences as subject-topic sentences. In G3, topic- change-marker overgeneralizations decline and children seem to have become proficient with the additive focus-particle. Instead, these children misinterpret over ¼ of all topic-change-markers as additives; though G4 could distinguish between both types of {-dA}. Given children's sensitivity to the distributional regularities in the input, if additive {-dA} is more frequent in child-directed speech than topic-change-marking {-dA}, older children, having outgrown their object-focus-bias, may overgeneralize the additive reading. We transcribed 3 episodes of a children's TV-show intended for the age-groups we tested. There were 164 {-dA}'s in this mini- corpus, 77 of which (47%) were additive focus-particles, while only 43 (26%) were topic-change-markers and 44 (27%) were not relevant for our test.1 As additive {-dA}'s are predominant in the input, children might misperceive additivity as the primary function of {-dA} and misanalyze some topic-change-markers as additives.
Keywords: Acquisition of Turkish, language development, additive focus, contrastive topic, prosody
Processing reflexives in L1 Turkish and L2 English: Evidence from eye-tracking
Münir Özturhan
Boğaziçi University
This study investigates antecedent preferences of Turkish-speaking learners of English in sentences such as ‘’Gordon realized that Richard had explained himself after arriving late for work.’’ According to Chomsky’s Binding Theory, Binding Principle A (BP-A) (1981), the reflexive himself must be bound by the local antecedent Richard but not by the non-local antecedent Gordon. Although this observation seems to be mainly true for English, certain reflexives such as Turkish kendi/kendisi do not appear to abide by BP-A (see Enç, 1989; Kornfilt, 1984; Meral, 2010; Palaz, 2013).Previous research on first language processing has also provided inconclusive evidence on syntactic biases in real-time antecedent retrieval for reflexives. Some studies found that native speakers’ antecedent preferences is syntactically constrained, conforming to the assumptions of BP-A (Nicol & Swinney, 1989; Sturt, 2003; Xiang et al., 2009); whereas others found that if the non-local antecedent has the same phi-features such as the same person, number and gender features with the reflexive and if this antecedent has discourse prominence, native speakers consider a non-local antecedent as a possible antecedent even at the early stages of processing (Badecker & Straub, 2002).Studies investigating second language (L2) processing of reflexives showed that L2 speakers’ initial antecedent preference was not syntactically constrained but it was influenced by discourse-related or pragmatic factors (Felser et al., 2009; Felser & Cunnings, 2012). This was true for learners with language backgrounds that behaved differently with respect to BT-A.The present study investigates whether Turkish-speaking learners of English will obey the assumptions of BP-A and choose only syntactically constrained antecedents in their second language or they will be influenced by other discourse-related features of the non-local antecedent. The motivation is even though English is a language that allows only local binding of reflexives; Turkish allows both local and non-local binding of reflexives. Results of a norming study showed that in neutral complex sentences in Turkish, native speakers of Turkish have a higher inclination to bind kendi and kendisi in Turkish with a non-local antecedent in the sentence.
The experimental items are taken from Sturt (2003) and adapted for the current experiment.
Sentences that are used exemplified in (1a, b) below:
(1a) Jonathan/Jennifer was pretty worried at the hospital. He/She remembered that the surgeon had pricked himself/herself with a used syringe needle. There should be an investigation soon. (Example from Experiment 1)
(1b) Jonathan/Jennifer was pretty worried at the hospital. The surgeon who treated Jonathan/Jennifer had pricked himself/herself with a used syringe needle. There should be an investigation soon. (Example from Experiment 2)
The predictions are:
These were tested via two eye-tracking experiments and a pen-and-paper questionnaire. Data were collected from 95 advanced learners of English. Experiment 1 was employed to 48 participants and Experiment 2 was employed to 47 participants. (Data are currently being analyzed.)After the main eye-tracking experiments all participants took an offline pen-and-paper questionnaire consisting of 24 experimental and 24 filler sentences to see non-native speakers’ final antecedent interpretations of reflexives. The results of the offline experiment showed that Turkish-speaking learners of English chose a syntactically-constrained accessible antecedent 79.30% of the trials whereas they preferred a matching-competitor inaccessible antecedent 20.70 of the trials for the referent of the reflexive.
Keywords: reflexive resolution, sentence processing, eye tracking
Münir Özturhan
Boğaziçi University
This study investigates antecedent preferences of Turkish-speaking learners of English in sentences such as ‘’Gordon realized that Richard had explained himself after arriving late for work.’’ According to Chomsky’s Binding Theory, Binding Principle A (BP-A) (1981), the reflexive himself must be bound by the local antecedent Richard but not by the non-local antecedent Gordon. Although this observation seems to be mainly true for English, certain reflexives such as Turkish kendi/kendisi do not appear to abide by BP-A (see Enç, 1989; Kornfilt, 1984; Meral, 2010; Palaz, 2013).Previous research on first language processing has also provided inconclusive evidence on syntactic biases in real-time antecedent retrieval for reflexives. Some studies found that native speakers’ antecedent preferences is syntactically constrained, conforming to the assumptions of BP-A (Nicol & Swinney, 1989; Sturt, 2003; Xiang et al., 2009); whereas others found that if the non-local antecedent has the same phi-features such as the same person, number and gender features with the reflexive and if this antecedent has discourse prominence, native speakers consider a non-local antecedent as a possible antecedent even at the early stages of processing (Badecker & Straub, 2002).Studies investigating second language (L2) processing of reflexives showed that L2 speakers’ initial antecedent preference was not syntactically constrained but it was influenced by discourse-related or pragmatic factors (Felser et al., 2009; Felser & Cunnings, 2012). This was true for learners with language backgrounds that behaved differently with respect to BT-A.The present study investigates whether Turkish-speaking learners of English will obey the assumptions of BP-A and choose only syntactically constrained antecedents in their second language or they will be influenced by other discourse-related features of the non-local antecedent. The motivation is even though English is a language that allows only local binding of reflexives; Turkish allows both local and non-local binding of reflexives. Results of a norming study showed that in neutral complex sentences in Turkish, native speakers of Turkish have a higher inclination to bind kendi and kendisi in Turkish with a non-local antecedent in the sentence.
The experimental items are taken from Sturt (2003) and adapted for the current experiment.
Sentences that are used exemplified in (1a, b) below:
(1a) Jonathan/Jennifer was pretty worried at the hospital. He/She remembered that the surgeon had pricked himself/herself with a used syringe needle. There should be an investigation soon. (Example from Experiment 1)
(1b) Jonathan/Jennifer was pretty worried at the hospital. The surgeon who treated Jonathan/Jennifer had pricked himself/herself with a used syringe needle. There should be an investigation soon. (Example from Experiment 2)
The predictions are:
- if Turkish learners of English show any early sensitivity to the gender of the accessible antecedent (the surgeon), the main effects of the non-local antecedent will not be observed at early eye-movement measures. It will be an indication of non-native speakers’ antecedent judgments are affected by the syntactic constraints; therefore, they would behave like the native speakers participating in the previous studies (Nicol & Swinney, 1989; Sturt, 2003; Xiang et al., 2009).
- if Turkish learners of English transfer long-distance binding feature of Turkish into English or are not affected by the gender manipulation at the early stages of processing, they will have a tendency to link the reflexive with the non-local antecedent. Additionally, if they show earlier sensitivity to the gender of the inaccessible antecedent (Jonathan/Jennifer) and are highly affected by the discourse-focus of the inaccessible antecedent, their antecedent search will be influenced by other cues to the interpretation such as lexical and semantic information. This will be an indication that there is a difference between native speakers and non-native speakers in terms of real-time antecedent retrieval for reflexives.
These were tested via two eye-tracking experiments and a pen-and-paper questionnaire. Data were collected from 95 advanced learners of English. Experiment 1 was employed to 48 participants and Experiment 2 was employed to 47 participants. (Data are currently being analyzed.)After the main eye-tracking experiments all participants took an offline pen-and-paper questionnaire consisting of 24 experimental and 24 filler sentences to see non-native speakers’ final antecedent interpretations of reflexives. The results of the offline experiment showed that Turkish-speaking learners of English chose a syntactically-constrained accessible antecedent 79.30% of the trials whereas they preferred a matching-competitor inaccessible antecedent 20.70 of the trials for the referent of the reflexive.
Keywords: reflexive resolution, sentence processing, eye tracking
Turkish-English speaking children’s bilingual emergent literacy skills following naturalistic experiences and a sound awareness intervention delivered in Turkish
Pınar Kolancalı, Sonali Nag
University of Oxford
The previous research has shown that explicit instruction in phonological awareness provides benefits for the development of emergent literacy skills, however the literature on the cross-linguistic transfer of these skills is limited. This paper presents the language and literacy outcomes in first and second language skills following naturalistic experiences and explicit instruction in phonological awareness (PA). The sample comes from a bilingual Kindergarten in Istanbul and the participants are five-year-old Turkish-English speaking bilingual children with Turkish as the dominant language (N = 21). The participants were first assessed for their Turkish and English PA, Turkish sentence repetition, Turkish letter knowledge and English expressive vocabulary prior to the intervention, to provide evidence of attainments from naturalistic experiences. The participants then received twelve 20-minute sessions of phonological games in Turkish focusing on the shared phonology of Turkish and English, after which they were assessed again. Information on each child’s home language and literacy environment was also collected with a questionnaire filled out by either mothers or fathers to assess the naturalistic experiences in the home. The results of the study reveals that mother education is positively correlated with children’s reading interest and emergent literacy skills, whereas father education is negatively correlated with these variables. The findings from the baseline and post-test language assessments show that Turkish emergent literacy skills are positively correlated with English language skills (all the correlation coefficient values are greater than .39). The post-test results also demonstrate significant results for the effectiveness of a short-term PA intervention for promoting gains in Turkish emergent literacy skills (p ≤ 0.001) and English PA (p ≤ 0.001). Taken together, these findings provide evidence for a cross-linguistic relationship within the bilingual child’s phonological domain. Importantly, this study provides preliminary evidence showing that improving of Turkish PA skills in five year olds benefits both Turkish and English learning.
Keywords: bilingualism, phonological awareness, literacy skills
Pınar Kolancalı, Sonali Nag
University of Oxford
The previous research has shown that explicit instruction in phonological awareness provides benefits for the development of emergent literacy skills, however the literature on the cross-linguistic transfer of these skills is limited. This paper presents the language and literacy outcomes in first and second language skills following naturalistic experiences and explicit instruction in phonological awareness (PA). The sample comes from a bilingual Kindergarten in Istanbul and the participants are five-year-old Turkish-English speaking bilingual children with Turkish as the dominant language (N = 21). The participants were first assessed for their Turkish and English PA, Turkish sentence repetition, Turkish letter knowledge and English expressive vocabulary prior to the intervention, to provide evidence of attainments from naturalistic experiences. The participants then received twelve 20-minute sessions of phonological games in Turkish focusing on the shared phonology of Turkish and English, after which they were assessed again. Information on each child’s home language and literacy environment was also collected with a questionnaire filled out by either mothers or fathers to assess the naturalistic experiences in the home. The results of the study reveals that mother education is positively correlated with children’s reading interest and emergent literacy skills, whereas father education is negatively correlated with these variables. The findings from the baseline and post-test language assessments show that Turkish emergent literacy skills are positively correlated with English language skills (all the correlation coefficient values are greater than .39). The post-test results also demonstrate significant results for the effectiveness of a short-term PA intervention for promoting gains in Turkish emergent literacy skills (p ≤ 0.001) and English PA (p ≤ 0.001). Taken together, these findings provide evidence for a cross-linguistic relationship within the bilingual child’s phonological domain. Importantly, this study provides preliminary evidence showing that improving of Turkish PA skills in five year olds benefits both Turkish and English learning.
Keywords: bilingualism, phonological awareness, literacy skills