1 of 29

�Complex syntax in Turkish-American returnees and L2 learners of Turkish residing in Turkey

Workshop on Third Factors in Language Design: The View from Heritage Languages

Inalco, Paris

June 6, 2025

Aylin Coşkun Kunduz�aylin.coskun-kunduz@uni-konstanz.de

Silvina Montrul�montrul@illinois.edu

Find more PowerPoint templates on prezentr.com!

2 of 29

Roadmap

Goal
Returnees
The Study

Relative clauses in Turkish
Research questions & Predictions
Methodology
Results

Sentence Repetition Task
Context-based Acceptability Judgement Task

Discussion & Conclusion

Find more PowerPoint templates on prezentr.com!

3 of 29

Goal

Second language (L2) acquisition

Late start (after puberty)
Limited classroom input

Is early age of exposure crucial for successful ultimate attainment?

Exposure to a language early on (before puberty) is important.

Exposure to a language early on (before puberty) is important but no guarantee of target-like command of the target language. Input also matters!

Heritage language (HL) acquisition

Early start (from birth)
Limited naturalistic input
Schooling in the majority language
No formal HL education

(Carreira & Kagan, 2011; Montrul, 2022)

(DeKeyser, 2000; Johnson & Newport, 1989)

Critical Period Hypothesis (CPH) postulates the existence of an optimal time frame in development during which humans are particularly sensitive to linguistic input.

(Abrahamsson & Hyltenstam, 2008, 2009; Lenneberg, 1967)

Find more PowerPoint templates on prezentr.com!

4 of 29

Immersion in HL/L2 in adulthood

In previous studies, HS and L2ers were tested in their dominant language environment, where their access to and use of the HL/L2 is limited.

What happens if HS and L2ers are tested in a naturalistic setting, when they are fully immersed in and using the HL/L2, in a majority language context?

Will the observed variability persist then?

Test returnees and L2ers residing in Turkey!

(Prela, Dąbrowska, & Llompart, 2024)

Find more PowerPoint templates on prezentr.com!

5 of 29

Returnees

L2 graph_3.png

(Montrul, 2015; Flores, 2020)

Figure 1. Relationship between amount of exposure to L1/L2 and age in typical monolinguals, heritage speakers, returnees and L2ers

Returnees are a particular case of HS, namely, bilinguals who grew up in a migration context and then returned, around or after puberty, to their (parents’) country of origin.

Find more PowerPoint templates on prezentr.com!

6 of 29

Changes in the former HL in returnees

Data comes from post-puberty Turkish-German returnees
Native-like performance in some structures but not in others
Task effects: Production ✅ vs. judgement ❌

Treffers-Daller, Özsoy and Hout (2007)

Complex embeddings in Turkish (noun clauses, relative clauses) using a picture description task
Participants:

Recent Turkish-German returnees (M_AoR = 16, M_LoR = 1-2 months)
Long-term returnees (M_AoR = 13, M_LoR = 8 years)

Findings:

HS & Recent returnees < Monolinguals = Long-term returnees

Conclusion: AoR or LoR effects?

Find more PowerPoint templates on prezentr.com!

7 of 29

The study

Find more PowerPoint templates on prezentr.com!

8 of 29

Novelty of the present study

AoR and LoR in Turkey are better controlled in the returnee group in this study.

Pre- and post-puberty returnees
Recent and long-term returnees

Returnees and L2ers in this study are from an English-speaking country.
Both production and judgement data are collected.
Multiple structures are being tested simultaneously to have a more comprehensive understanding of returnee & L2 grammars:

Relative clauses
Verbal passives
Binding/Anaphoric dependencies
Differential Object Marking and Evidentiality (not reported here)

Find more PowerPoint templates on prezentr.com!

9 of 29

Relative clauses in Turkish

Turkish RCs are prenominal, and there is no overt relative pronoun.

OVS word order

SVO word order

Subject relative clause (SRC)
(1) a. [__ kız-ı öp-en] adam � girl-ACC kiss-SRC man � ‘The man that is kissing/kissed the girl’
Object relative clause (ORC)
b. [kız-ın __ öp-tüğ-ü] adam � girl-GEN kiss-ORC-POSS.3SG man � ‘The man that the girl is kissing/kissed’

(Göksel & Kerslake, 2005)

Find more PowerPoint templates on prezentr.com!

10 of 29

Acquisition of relative clauses in Turkish

RCs are a late accomplishment, and adult-like acquisition of RCs does not occur until the age of 6-7 in monolinguals.

ORCs are particularly challenging.

RCs are vulnerable in Turkish HS and L2ers as well.

In Turkish (long-term) returnees, RCs seem to be native-like.

(Özcan, 1997; Slobin, 1986)

(Altan, 2016; Coşkun Kunduz & Montrul, 2022)

(Antonova-Unlu et al., 2021)

Find more PowerPoint templates on prezentr.com!

11 of 29

Research questions and predictions

To what extent do Turkish HS returnees and L2ers show convergence towards target-like competence of RCs upon full immersion in Turkish after their return?

Early age of acquisition

Returnee�L2

Maximal input (timing does NOT matter)

Returnee�L2

Early age of acquisition +
Maximal input (timing DOES matter; before puberty)

Returnee�L2

CPH✓

Find more PowerPoint templates on prezentr.com!

12 of 29

Research questions and predictions

Which areas of the grammar can be robustly acquired with maximal input and which ones may be subject to maturational effects past puberty?

Particularly challenging: ORCs

To what extent do the situational factors (e.g., age of return to Turkey, length of residence in Turkey, the status of and contact with the former dominant language) play a role in HL development in returnees?

The earlier the AoR to Turkey and the longer LoR in Turkey are, the better the returnees and L2ers may perform.
English as a lingua franca may negatively affect the degree and speed of the further development of Turkish in returnees and L2ers.

(Flores & Snape, 2021; Tomiyama, 2009 among others)

Find more PowerPoint templates on prezentr.com!

English as a lingua franca may negatively affect the degree and speed of the reactivation of Turkish.

For the degree of contact with English after return, previous studies have shown that those who returned from English-speaking countries were better at maintaining their knowledge of English even after years of full immersion in the former HL environment thanks to the status of English as a lingua franca (Flores & Snape, 2021; Tomiyama, 2009 among others). Therefore, Turkish-American returnees in this study may continue to have contact with English on a daily basis through social media, school, songs or films, and this may negatively affect their degree and speed of development in Turkish.

English as a lingua franca may negatively affect the degree and speed of the reactivation of Turkish.

13 of 29

Participants

30 Turkish monolinguals
30 Turkish second-generation adult HS in the US
34 Turkish-American returnees
24 English-L1 Turkish-L2ers

Figure 2. Relation between AoR to Turkey and LoR in Turkey after return in returnees and L2ers

		Returnee	L2
AoR	Mean	13.5	31.7
	SD	8	12
	range	4–34	18–68
LoR (years)	Mean	14	15
	SD	10.6	8
	range	1–50	1–34
Contact with English	Mean	37%	65%
	SD	16	18
	range	10–75	20–99

Find more PowerPoint templates on prezentr.com!

14 of 29

1. Sentence repetition (Srep) task

Participants are asked to repeat sentences verbatim.
A total of 118 items:

48 RCs (24 ORCs + 24 SRCs)
48 long verbal passives with by-phrase (24 agent-patient + 24 experiencer-theme verbs)
24 simple active SOV + 12 fillers
2 practice items

(Marinis & Armon-Lotem, 2014; Topbaş et al., 2013)

Figure 3. Sample stimuli from the SrepT

Find more PowerPoint templates on prezentr.com!

15 of 29

128 items testing RCs with 8 conditions:

2 RC types (SRC, ORC)
2 roles within matrix clause (subject, object)
2 context types (SRC, ORC)

16 randomized blocks; 4 lists
Each participant saw a total of 32 RCs.
Bimodal presentation

On a nice Sunday, a woman was running. On the way, she saw an old man and a little boy. They were chatting. And then the man greeted the woman.

Kadını selamlayan adam çocukla konuştu.�‘The man that greeted the woman chatted with the boy.’

A sample item testing RC: �(SRC in subject position)

(Gürel, 2002)

2. Context-based AJT

EVET

HAYIR

Find more PowerPoint templates on prezentr.com!

16 of 29

Results

Find more PowerPoint templates on prezentr.com!

17 of 29

Results: SrepT

ORC < SRC in all groups
L2 < HS < Returnees = Mono.

Early age of exposure is important but not sufficient by itself for successful acquisition.

Input also matters!

Figure 4. Mean accuracy percentages by group and structure

in the Srep task

Figure 5. Individual variation in the SrepT

Find more PowerPoint templates on prezentr.com!

18 of 29

Results: Srep task

-nin

Example (Omission):

çocuklar kuyuya düşmüşler� kids well.DAT fell�‘The children who pushed you fell into the well.’

-GEN

Seni it-en

you.ACC push-SRC

Find more PowerPoint templates on prezentr.com!

19 of 29

Results: AJT

ORC = SRC in all groups
HS = L2 < Returnees = Mono.�(p = .0524)

Early age of exposure is important but not sufficient by itself for successful acquisition.

Input also matters!

Figure 6. Mean accuracy percentages by group and structure

in the AJT

Figure 7. Individual variation in the AJT

Find more PowerPoint templates on prezentr.com!

20 of 29

Results

Earlier AoR to Turkey 🡪 better RC production in L2ers

Figures 8 & 9. Correlations between mean accuracy percentages and AoR and LoR to Turkey in returnees in the SrepT (production) and in the AJT (judgement)

Earlier AoR to Turkey 🡪 better RC judgement in both groups

Find more PowerPoint templates on prezentr.com!

21 of 29

Discussion

To what extent do Turkish HS returnees and L2ers show convergence towards target-like competence of RCs upon full immersion in Turkish after their return?

Early age of acquisition

Returnee�L2

Maximal input (timing does NOT matter)

Returnee�L2

Early age of acquisition +
Maximal input (timing DOES matter; before puberty)

Returnee�L2

CPH✓

Find more PowerPoint templates on prezentr.com!

22 of 29

Discussion

Which areas of the grammar can be robustly acquired with maximal input and which ones may be subject to maturational effects past puberty?

Particularly challenging: ORCs ﬩
No subject advantage in comprehension 🡪 In line with O’Grady (2011)’s emergentist approach

SRCs are more frequent & prominent, but linear distance favors ORCs

SRC advantage in production 🡪 In line with previous research

To what extent do the situational factors (e.g., age of return to Turkey, length of residence in Turkey, the status of and contact with the former dominant language) play a role in HL development in returnees?

Earlier AoR to Turkey 🡪 better judgements in both returnees & L2ers� better production in L2ers

(Avram et al., 2023; Coşkun Kunduz & Montrul, 2022)

Find more PowerPoint templates on prezentr.com!

23 of 29

Conclusion

Complex morphosyntax in Turkish is still nimble post-puberty!

Early age of acquisition is crucial but not sufficient for successful ultimate attainment.

Input also matters!

But timing of input does NOT matter (before or after puberty).

Find more PowerPoint templates on prezentr.com!

24 of 29

THANK YOU!

Find more PowerPoint templates on prezentr.com!

25 of 29

Selected bibliography

Abrahamsson, N., & Hyltenstam, K. (2009). Age of onset and nativelikeness in a second language: Listener perception versus linguistic scrutiny. Language Learning, 59(2), 249–306.

Altan, A. (2016). Relative Clauses in L2 Turkish. Mersin Üniversitesi Dil Ve Edebiyat Dergisi, 13(1), 1-37.

Coşkun Kunduz, A., & Montrul, S. (2022). Relative clauses in child heritage speakers of Turkish in the United States. Linguistic Approaches to Bilingualism. https://doi.org/10.1075/lab.21027.cos

Flores, C. (2020). Attrition and reactivation of a childhood language: The case of returnee heritage speakers. Language Learning, 70, 85-121.

Montrul, S. (2022). Native Speakers Interrupted. Cambridge, UK: Cambridge University Press.

Treffers-Daller, J., Özsoy, A. S., & van Hout, R. (2007). (In)Complete acquisition of Turkish among Turkish–German bilinguals in Germany and Turkey: An analysis of complex embeddings in narratives. International Journal of Bilingual Education and Bilingualism, 10(3), 248–276.

Find more PowerPoint templates on prezentr.com!

26 of 29

Funding

NSF Doctoral Dissertation Research Improvement Grant
Graduate College Dissertation Completion Fellowship
School of Literatures, Cultures & Linguistics Dissertation Completion Fellowship (*declined due to fellowship conflict)
D.A. Kibbee Prize for SLCL Dissertation Completion Fellowship recipient judged as ‘most outstanding’ (*declined due to fellowship conflict)
Alexander von Humboldt Fellowship

Find more PowerPoint templates on prezentr.com!

27 of 29

�Complex syntax in Turkish-American returnees and L2 learners of Turkish residing in Turkey

Workshop on Third Factors in Language Design: The View from Heritage Languages

Inalco, Paris

June 6, 2025

Aylin Coşkun Kunduz�aylin.coskun-kunduz@uni-konstanz.de

Silvina Montrul�montrul@illinois.edu

Find more PowerPoint templates on prezentr.com!

28 of 29

1. Sentence repetition (Srep) task

ORC (object/center-embedded)	ORC (subject/left-branching)
Çocuklar [iç-tik-leri çorba]-yı sevdiler children eat-ORC-POSS.3PL soup-ACC liked ‘The children liked the soup that they ate.’	[Çiftçi-nin besle-diğ]-i at onu ısırdı farmer-GEN eat-ORC-POSS.3SG horse him bit ‘The horse that the farmer is feeding bit him.’
SRC (object/center-embedded)	SRC (subject/left-branching)
Keçi [maymun-u sev-en kız]-a bakıyordu goat monkey-ACC like-SRC girl-DAT looked ‘The goat was looking at the girl that pat the monkey.'	[Kuş-u yakala-yan tilki] kedi-ye kızmıştı. bird-ACC catch-SRC fox cat-DAT got.angry ‘The fox that caught the bird got angry with the cat.’

Find more PowerPoint templates on prezentr.com!

29 of 29

	Conditions	Number of token sets	Number of YES and NO conditions
RCs	8 conditions = 2 RC type * 2 role within matrix clause * 2 context type (SRC, ORC)	16	2 YES; 2 NO
Verbal passives	8 conditions = 2 verb type (agent-patient vs. experiencer-theme) * 2 verb form (active vs. passive) * 2 context type	16	4 YES; 4 NO
Anaphoric dependencies	8 conditions = 4 pronoun/anaphor type (o vs. null, kendisi and kendi) * 2 context type (bound vs disjoint)	16	10 YES; 2 NO

2. Context-based AJT

Table 2. Stimuli in the AJT

Find more PowerPoint templates on prezentr.com!