Language is a living, dynamic entity that constantly evolves to meet the social needs of its speakers. For bilingual individuals, these social requirements can vary significantly depending on their linguistic environment. When social shifts and changes in language contexts lead to reduced use and exposure of the first language in an immersive second-language environment, the resulting changes in the native language and its usage can be referred to as first language attrition. These changes can be observed, for example, among emigrants who have left their native country and lived for an extended period in a second-language environment.

Changes in the native language and its use are frequently observed at the lexical level, both in informal conversations and formal lexical tasks. Verbal Fluency (VF) tasks are a versatile tool for investigating lexical retrieval. Language attrition research shows that immersed bilinguals tend to produce fewer words in VF tasks compared to monolinguals. However, the underlying factors that hinder optimal performance have not been systematically identified, and the value of these tasks as a research tool in language attrition studies has been questioned. While VF total scores offer a general measure of task performance, they do not provide insight into the processing strategies underlying lexical retrieval, such as frequency and type of errors, the temporal distribution of words, and the clustering and switching patterns employed during VF task performance.

This dissertation by publication consists of three studies. The overarching aim of the dissertation is to identify processes underlying VF task performance in a group of immersed Finnish-English bilinguals living in Northern California. The first study (Study I) introduced and validated a systematic method for comprehensive analysis of VF task performance, including total scores, errors, temporal parameters, and clustering and switching strategies. The subsequent articles (Study II and Study III) applied this analysis method to VF task data from immersed bilinguals living in Northern California, US in the native language, Finnish (L1), and their second language, English (L2). Task performance in L1 was also compared to that of a control group of Finnish speakers living in Finland. Two extralinguistic variables were included to consider the impact of the shared language background on VF task performance in the immersed bilingual group: duration of immersion in the L2 environment and the frequency of L1 use during this time. Studies II and III demonstrate that the analysis methods introduced in Study I can effectively identify similarities and differences in language processing strategies employed during VF task performance.

Immersed bilinguals performed similarly in L1 and L2 VF tasks. Study II demonstrated that more frequent L1 use was associated with a higher total score, and this positive influence extended to both languages. These findings suggest that the observed strategies utilized in VF task performance are not limited to L1 but reflect general language processing strategies within the immersed bilingual group, emphasizing the importance of including both languages of immersed bilingual speakers in language attrition studies.

Differences in L1 performance between the immersed bilingual group and the control group living in Finland were detected in Studies II and III, suggesting potential cognitive adaptations in the group of immersed bilinguals. Immersed bilinguals utilized rapid lexical retrieval more efficiently than the control group. They also relied more systematically on clustering to reach a higher total score, while appearing to lack flexibility in switching between clusters once a cluster was exhausted. This could indicate that immersed bilinguals rapidly generated words belonging to a single cluster at the beginning of the task (e.g., cluster “pets” in the category “animals”) but were less successful in switching to a new category or in returning to a previous cluster (e.g., “pets” - “jungle animals” - “pets”), resulting in reduced overall performance. The findings from this dissertation offer cautious optimism that the proposed method could identify group specific characteristics of lexical retrieval in bilingual groups immersed in a second language environment.

In conclusion, this dissertation demonstrates that a detailed analysis of VF task performance can enhance our understanding of the processes underlying lexical retrieval in varied bilingual groups. In addition to language attrition studies, the method proposed in Study I could be applied across diverse populations and settings. Considering that VF tasks are among the most common tasks for evaluating lexical retrieval in neuropsychological language assessments, the application of the proposed analysis methods extends to various research fields, including, but not limited to, speech pathology, neuropsychology, and linguistics