Semantic verbal fluency in native speakers of Turkish: a systematic review of category use, scoring metrics and normative data in healthy individuals

ABSTRACT

Introduction

Semantic verbal fluency (SVF) is a widely used measure of frontal executive function and access to semantic memory. SVF scoring metrics include the number of unique words generated, perseverations, intrusions, semantic cluster size and switching between clusters, and scores vary depending on the language the test is administered in. In this paper, we review the existing normative data for Turkish, the main metrics used for scoring SVF data in Turkish, and the most frequently used categories.

Method

We conducted a systematic review of peer-reviewed papers using Medline, EMBASE, PsycInfo, Web of Science, and two Turkish databases, TR-Dizin and Yok-Tez. Included papers contained data on the SVF performance of healthy adult native speakers of Turkish, and reported the categories used. Versions of the SVF that required participants to alternate categories were excluded. We extracted and tabulated demographics, descriptions of groups, metrics used, categories used, and sources of normative data. Studies were assessed for level of detail in reporting findings.

Results

1400 studies were retrieved. After deduplication, abstract, full text screening, and merging of theses with their published versions, 121 studies were included. 114 studies used the semantic category “animal”, followed by first names (N = 14, 12%). All studies reported word count. More complex measures were rare (perseverations: N = 12, 10%, clustering and switching: N = 5, 4%). Four of seven normative studies reported only word count, two also measured perseverations, and one reported category violations and perseverations. Two normative studies were published in English.

Conclusions

There is a lack of normative Turkish SVF data with more complex metrics, such as clustering and switching, and a lack of normative data published in English. Given the size of the Turkish diaspora, normative SVF data should include monolingual and bilingual speakers. Limitations include a restriction to key English and Turkish databases.

KEYWORDS:

• Semantic verbal fluency
• clustering
• switching
• Turkish
• semantic categories

Introduction

Semantic verbal fluency (SVF) is a widely used neuropsychological assessment to examine frontal executive functions (EF) and access to semantic memory (Patterson, 2011). For SVF, participants are asked to produce as many words of a given category as possible in a short amount of time, usually 60 seconds. The most frequently used category is animals, but other categories such as supermarket items or fruit and vegetables have also been used. SVF is often paired with another type of verbal fluency test, phonemic or letter fluency. In letter fluency, the participant is asked to produce as many words as possible that start with a specific letter. In English, these letters are usually F, A, and S or C, F, and L (Amunts et al., 2020).

SVF is part of several standard cognitive assessment batteries, such as the Addenbrooke’s Cognitive Examination (Mathuranath et al., 2000) or Delis-Kaplan Executive Function System (D-KEFS) (Delis et al., 2001). Many conditions such as Alzheimer’s Disease (Gomez & White, 2006; Monsch et al., 2020), Parkinson’s Disease (Piatt et al., 2010; Raskin et al., 1992) and traumatic brain injury (Kavé et al., 2011; Raskin & Rearick, 1996; Zakzanis et al., 2011) consistently affect participants’ SVF performance.

The most commonly used scoring metric for SVF is the total number of correct unique words produced within the time-limit (Ardila et al., 2006). However, SVF is considered a multifactorial task and simply considering the total number of correct words is not thought to be sufficient to fully capture a participant’s performance. Some researchers also report the number of words produced in the first, second, third, and fourth quarter of the 60 second time-frame. Errors can provide a qualitative measure of SVF performance, and are divided into two subtypes: intrusions and perseverations (Raboutet et al., 2010). Intrusions or category violations are additional words that do not belong to the desired category (Raskin & Rearick, 1996). Perseveration refers to the repetition of words and is divided into three types (Pekkala et al., 2008; Sandson & Albert, 1987): recurrent (repeated words separated by other words, e.g., cat, dog, fox, cat), stuck-in-set (repetition of words from a previous category, such as animals, in the current category, such as supermarket items), and continuous (repeated production of a single word, e.g., cat, cat, cat). Generally, all repeated words tend to be classified as perseverations (Galaverna et al., 2016; Henry & Phillips, 2006), regardless of the location of the words produced or their semantic integrity.

Other SVF scoring metrics focus on uncovering the process by which the sequences are generated. Semantic clustering is a technique of grouping words that are similar in meaning and belong to predefined subgroups (Ober et al., 1986; Raskin et al., 1992; Robert et al., 1998). Troyer et al. (1997) proposed a formal analysis procedure that systematically identifies clusters of related words and assesses the number of switches between clusters as well as cluster size to allow for more fine grained analyses. For example, older people produce fewer words and switches than young people, but create larger clusters (Troyer, 2000; Troyer et al., 1997). While people with dementia normally produce fewer words than healthy controls, switching is impaired in frontal lobe related neurodegenerative diseases (e.g., Parkinson’s disease) and cluster size decreases in temporal lobe involved neurodegenerative diseases (e.g., Alzheimer’s disease) (Troyer, Moscovitch, Winocur, Alexander, et al., 1998; Troyer, Moscovitch, Winocur, Leach, et al., 1998). Troyer’s approach has been adapted widely to assess these scoring metrics, which are both thought to be necessary for successful SVF performance (Hurks et al., 2010; Lanting et al., 2009; Weiss et al., 2006; Zhao et al., 2013).

Given that SVF is one of the most widely used neuropsychological tests, normative data have been collected to establish expected SVF scores for healthy native speakers in several different languages. Examples include English (Tombaugh, 1999), Spanish (Benito-Cuadrado et al., 2002), Arabic (Khalil, 2010), Chinese (Feng et al., 2012), Dutch (Van Der Elst et al., 2006), Hebrew (Kavé, 2005), Greek (Kosmidis et al., 2004), and Persian (Ghasemian-Shirvan et al., 2018). Normative studies will typically ensure that variation in performance according to age, gender and education is already accounted for in the published norms. Collecting normative data for different languages/ethnic groups is considered necessary given that SVF scores have been found to vary depending on the language/ethnic group the task is administered. For example, Kempler et al. (1998) found that Hispanic individuals produced significantly fewer animal names than Chinese, White, and Vietnamese groups and that Vietnamese individuals produced more animal names than Chinese, White, and Hispanic groups. These results have been linked to the length of the animal names spoken in each language (e.g., the word “dog” in Spanish is perro, while in Vietnamese is chó); longer animal names result in fewer words generated. Therefore, it is important to establish normative data for the variations of SVF in different languages, determine what metrics have been used, and establish whether additional normative data or additional scoring methods for existing normative data would be beneficial. Both demographic information and cultural background can impact semantic memory organization, category size and content (Rosselli et al., 2002; Strauss et al., 2006). Therefore, semantic categories included in SVF require scoring guidelines (Olabarrieta-Landa et al., 2017) and revealing culturally and linguistically unique retrieval strategies requires in-depth study of data from the population being considered (Olabarrieta-Landa et al., 2017). In this review, we focus on the Turkish language. Turkish is predominantly spoken by around 84 million people in Turkey (TUIK, Turkish Statistical Institute, 2022). There is also a large Turkish diaspora of more than 6.5 million all over the world, 5.5 million of whom live in Western Europe (Turkish Ministry of Foreign Affairs, Turkish, 2022). Turkish differs substantially in word structure from more frequently studied languages, such as English. Turkish is a member of the Turkic language family, which also includes Azerbaijani and Turkmen. Its morphology is agglutinative (Durrant, 2013). Information such as case of nouns, plural/singular, or tense is indicated by adding an affix to a root word (Istek & Cicekli, 2007).

Since the design of our systematic review was descriptive, we did not have any a priori hypotheses. The systematic review addresses three research questions:

RQ1:

What SVF tasks are commonly used for Turkish?

RQ2:

What are the most commonly used scoring metrics? To what extent are more complex scoring approaches, such as clustering and switching (Troyer, 2000), used?

RQ3:

What normative studies exist for Turkish, and how do the tasks and metrics map onto the existing literature on SVF in Turkish?

Materials and methods

Search strategy

This review was registered on Prospero with the protocol number CRD42020201585 and passed the suitability examination on 27 October 2020. In our study, we used the PICOS structure to frame the review questions, which is one of the most widely used frameworks proposed by Richardson et al. (1995) to help evaluate the validity, applicability and completeness of a systematic review. The PICOS framework creates a question that identifies: (1) the Patient problem or Population, (2) the Intervention, (3) the Comparison (if there is one), (4) the Outcome(s), and (5) the Study type. The PICOS structure of the study is summarized in Table 1. We searched Web of Science Core Collection, PsycINFO, EMBASE, MEDLINE, and two Turkish databases, TR-Dizin (National Database Index by ULAKBIM, the Turkish Academic Network and Information Centre) and Yok-Tez (Databases of National Thesis Center of the Council of Higher Education).

Table 1. PICOS specification of the systematic review inclusion criteria.

Term	Description
Population	Healthy native speakers of Turkish aged 16 and over.
Intervention	Standard semantic verbal fluency test. All semantic categories are included.
Comparison	In addition to normative data, we include papers that compare groups of healthy native speakers and papers that compare healthy native speakers to people with a health condition, such as neurodegenerative diseases or a mental disorder. Studies of bilingual speakers should have a monolingual control group.
Outcome	Scoring metrics used for semantic verbal fluency.
Study	Experimental studies that include more than one participant.

The general Boolean search formula was ((semantic fluency OR category fluency OR verbal fluency OR semantic verbal fluency OR COWAT OR Controlled Oral Word Association Test OR animal fluency OR animal naming) AND turk*), applied to the Topic, Title, Abstract, Author and Keywords fields. Local database searches were performed first using the English version of the formula, then using the Turkish translation, removing the *turk at the end. English keywords and their Turkish equivalents are given in Table 2. An initial search was performed on 5 October 2020 and updated using the same search criteria for works that were published after our initial search (9 August 2023). The search was conducted without any start date restrictions. While the Controlled Oral Word Association Test (COWAT, (Rodríguez-Aranda & Martinussen, 2006)) assesses letter fluency, we included it as a search term because some studies use COWAT to refer to semantic verbal fluency (Ersan, 2014; Uslu, 2012) or as an umbrella term for verbal fluency (Kiraç et al., 2014).

Table 2. Turkish translations of English query terms for search of Turkish databases.

English	Turkish
Semantic verbal fluency	Semantik sözel akıcılık
Semantic fluency	Semantik akıcılık
Verbal fluency	Sözel akıcılık
Category fluency	Kategori akıcılık
Animal fluency	Hayvan akıcılığı
Animal naming	Hayvan sayma
Controlled Oral Word Association Test (COWAT)	Kontrollü Kelime Çağrışım Testi

Inclusion and exclusion criteria

We included studies that reported primary SVF data from a group of at least two native Turkish speakers, and that were available in digital form. SVF data should be collected using the standard paradigm, i.e., naming as many words as possible from a given category in a short time, typically 60 seconds. At least one group of participants should consist of healthy adults.

The exclusion criteria fell into three broad categories: population; methodology; and accessibility.

The population-related exclusion criteria were: “study of children”; “no healthy control group”; “no monolingual control group”; and “potential effect on SVF, no control”. We defined “study of children” as a study that only reported data from participants under the age of 16. However, studies that specified a minimum age of 15 or 16 years for their participants, and that reported an average age of 18 or older, were included (Bora et al., 2019; Yilmaz, 2014). Studies that only focused on people with medical conditions (“no healthy control group”) or only on bilingual people (“no monolingual control group”) were also excluded. Finally, we excluded studies that reported individuals with potential undiagnosed conditions that might affect their performance on the SVF such as coronary artery disease (“potential effect on SVF, no control”) (Ünlü et al., 2013).

The methodology-related exclusion criteria were: “only letter fluency”; “composite results”; and “no results reported”. Studies that performed only letter fluency, phonemic fluency, or COWAT tests without any SVF component, were excluded as “only letter fluency”. Studies reporting “composite results” did not provide metrics on SVF performance for a specific category. Instead, scores were merged across SVF categories (Aydınoğlu, 2015; Erol et al., 2012; İpekten, 2018; Kaya & Zengin Alpözgen, 2022), participants completed tests that involve alternating categories (Erdogan, 2016), scores from letter fluency and SVF were mixed (Beşer, 2019; Gultekin et al., 2017; Kılavuz Ören, 2020), or the type of verbal fluency test was not specified (Cevik et al., 2016; Gücüyener et al., 1998; Kandemir et al., 2009; Karahan et al., 2021). If studies reported no verbal fluency results at all, they were classified as “no results reported” (Akgün, 2010; Boyle et al., 2021; Er, 2014; Midi et al., 2011).

The accessibility-related criteria were “abstract only”, which means that no full paper was available, and “no online version available”, which applied to studies, in particular Turkish theses, that were only available in printed form.

Reference management

Database search results were exported as BibTeX or RIS format except for Yok-Tez, which contains all Master, PhD and medical proficiency degree theses for diploma accreditation in Turkey. Although this system allows searching and downloading articles, it is not possible to save search results as bibliographic information. Therefore, the first author manually input references for the studies downloaded from Yok-Tez into Mendeley. Screening and data extraction were conducted using Covidence (2013). Duplicates were initially resolved automatically by Covidence. Remaining duplicates were resolved manually by prioritizing the source in the order of MEDLINE, Embase, PsycInfo, Web of Science.

Screening and extraction

All studies were screened by two members of the review team (RYK, KK, MKW, SM) in the abstract and full text screening stages. Studies that were only available in Turkish were reviewed by RYK and KK, who are native speakers of Turkish, while studies available in English were reviewed by RYK, MKW, and SM. Reviewers used the note facility to highlight potential issues and record their reasoning behind the decisions made. Disagreements were resolved through discussion between members of the review team.

Nine publications (Çabuk et al., 2020; Demiray & Ertan, 2023; Kiraç et al., 2014; Mutlu et al., 2021; Özcan et al., 2016; Özçelik-Eroğlu et al., 2014; Sezikli et al., 2018; Töret & Özdemir, 2021; Uzgan et al., 2021) are theses that were later published as a peer-reviewed article; for those studies, we only extracted data from the published version, which is more likely to be accessible to the international community.

The PRISMA diagram (Figure 1) documents the screening process. The final number of studies analyzed is 121.

Figure 1. PRISMA flow chart.

The reviewers jointly designed an extraction template that was designed with the research questions in mind. Extraction was done by RYK, supported by KK. SM and MKW were consulted in case of questions.

For each study, we extracted three types of general information: Language of the study (Turkish/English); Type of publication (thesis/journal article etc.); and Country in which the study was conducted (Turkey/Other). In order to establish commonly used tasks and scoring metrics (RQs 1 and 2), we extracted all semantic categories used (e.g., animals, vegetables), the word count metrics reported (mean/minimum/maximum/standard deviation etc.), any alternative analysis methods (clustering and switching/computational etc.), and the administration time of the SVF task. In order to distinguish between normative and non-normative studies (RQ3), we extracted information about study type. First, we determined the study design (longitudinal versus cross-sectional), and whether SVF was the focus of the study or part of a general assessment. If the study included a patient sample, we noted the type of health condition (e.g., cognitive impairment or dementia/mental illness/Parkinson’s Disease etc.). If the study compared two or more groups of healthy participants, we noted how the groups were defined. Finally, we extracted demographic characteristics for the healthy native speakers of Turkish who were assessed. This included the age of the participants (older adults, aged $\ge$64, adults aged 18–64, young people aged 16–25, and children/pediatric groups ($\le$16), and whether multiple age groups were differentiated. We noted the number of healthy participants, whether gender information was recorded, whether an indicator of level of education was used (years/levels/none), whether socioeconomic status was reported (No/Yes, occupation/Yes, income bracket), and age groups reported.

In the main results tables, we report the data extracted from each study and highlight missing information. The tables are summarized in the text using counts and percentages. For the tables, studies were split into normative and non-normative (c.f. Table 3). Non-normative studies were additionally split into studies comparing groups of healthy adults (c.f. Table 4), studies comparing healthy controls to people with a mental health condition (c.f. Table 5), studies comparing healthy controls to people with a neurodegenerative disease (c.f. Table 6), and studies comparing healthy adult controls to people with other diseases (c.f. Table 7).

Table 3. List of normative studies with full extracted data. [*]Because our scope is adults, the results of the 30 children used in the study were excluded. Therefore, our assessment is based on 120 adults and the total number of people is 30 less than the actual number reported in the study.

General Information			Demographics				Semantic Verbal Fluency		Analysis
Study ID	Language/ Publication type/ Country	Is SVF Focus?	Participant (N) Gender (M/F)	Age groups N (M/F)	Education groups N (M/F)	Economic status	Categories	Duration	Word count metrics	Other metrics	Statistical scales
Tumaç (1997)	Turkish/Thesis/Turkey	Yes	180 (90/90)	15–28 ages: 60 32–45 ages: 60 50–75 ages: 60	0–5 years: 60 6–12 years: 60 12+ years: 60	Occupation	Animals	60 sec.	Mean, SD	Perseveration	Total word count, F-value
Tuncer (2012)	Turkish/Thesis/Turkey	Yes	382 (170/ 212)	18–24 ages: 78 25–34 ages: 89 35–44 ages: 36 45–54 ages: 51 55–64 ages: 64 65+ ages: 67	illiterate: 19 1–8 years: 87 9–11 years: 113 12+ years: 163	N/A	Animals, Vegetables and Fruits, Vehicles, Clothes, Parts of the Body, Furniture	60 sec.	Mean, SD	N/A	Total word count, chi-square, F-value, t-value
Özdemir (2015)	Turkish/Thesis/Turkey	Yes	120* (60/60)	18–29 ages: 30 30–44 ages: 30 45–59 ages: 30 60+ ages: 30	1–8 years: (20/20) 9–11 years: (20/20) 12+ years: (20/20)	N/A	Breakfast Items, Famous People, Food, Beverages, Household Items	90 sec.	Mean, SD, min, max	N/A	Total word count, chi-square, U-value
Şentürk (2019)	Turkish/Thesis/Turkey	Yes	200 (99/101)	18–29 ages: 71 30–39 ages: 60 40–49 ages: 69	5–8 years: 60 9–11 years: 66 12+ years: 74	N/A	Animals	60 sec.	Mean, SD, median, 5%ile, 95%ile	N/A	Total word count, percentiles, F-value
Özdemir and Tunçer (2021)	Turkish/Article/Turkey	Yes	58 (30/28)	60–81 ages: 58	5–8 years: 20 9–11 years: 21 12+ years: 17	N/A	Breakfast Items, Famous People, Food, Beverages, Household Items	60 sec.	Mean, SD, quarters (each 15 sec.)	N/A	Total word count, chisquare, z-score, F-value
Aki et al. (2022)	English/Article/Turkey	Yes	415 (208/207)	15–24 ages: (46/42) 25–34 ages: (36/33) 35–44 ages: (30/36) 45–54 ages: (35/32) 55–64 ages: (31/34) 65+ ages: (30/30)	5–8 years: (61/66) 9–11 years: (66/62) 12+ years: (81/79)	N/A	Animals, Human Names	60 sec.	Mean, SD	Perseveration	Total word count, chi-square, F-value
İlkmen and Büyükişcan (2022)	English/Article/Turkey	Yes	1431 (727/704)	18–19 ages: (60/60) 20–29 ages: (102/104) 30–39 ages: (103/103) 40–49 ages: (106/104) 50–59 ages: (105/102) 60–69 ages: (102/100) 70–79 ages: (92/84) 80–89 ages: (54/47)	0–5 years: (159/160) 6–12 years: (205/209) 12+ years: (212/199)	N/A	Animals	60 sec.	Mean, SD, quarters (each 15 sec.)	Perseveration, category violations	Total word count, percentiles, F-value, t-value

Table 4. List of studies that compare groups of healthy native speakers with full extracted data.

General Information			Demographics				Semantic Verbal Fluency		Analysis
Study ID	Language/ Publication type/ Country	Is SVF focus?	Participant N (M/F)	Age Min-Max (SD)	Education years Min-Max (SD)	Economic status	Categories	Duration	Word count metrics	Other metrics	Group comparison	Topics
Yazici (2019)	Turkish/ Thesis/ Turkey	No	160 (80/80)	20–54	Min: 9	N/A	Animals	Unknown	Mean, SD	Perseveration, category violations	Monolingual (Turkish) Bilingual (Kurdish and Turkish)	Bilingualism
Gökçe et al. (2021)	English/ Journal Article/ Turkey	No	48 (25/23)	18–50 (11.10)	Min: 15	N/A	Supermarket Items	60 sec.	Median, min, max	N/A	Tennis players, sedentary people	Physical activity
Gökçe (2020)	Turkish/ Thesis/ Turkey	No	54 (25/29)	18–25	Mean: 13.50	N/A	Animals, Vegetables and Fruits, Supermarket Items	60 sec.	Mean, SD	N/A	Three exercise groups: fencers, swimmers, sedentary focusing on gender differences
Yeniçeri (2019)	Turkish/ Thesis/ Turkey	No	50 (24/26)	18–30	N/A	N/A	Animals	60 sec.	Mean, SD	N/A	Cognitive functions with and without balance exercises
Ekin and Çebi (2021)	Turkish/ Journal Article/ Turkey	No	80 (40/40)	60–80 (3.92)	Min: 8	Occupation	Animals	60 sec.	Mean, SD, median, min, max	N/A	Neuropsychological tests, demographics and reading habits	Non-physical leisure activity
Yıldırım and Ogel-Balaban (2021)	English/ Journal Article/ Turkey	No	70 (36/34)	55–84 (8.59)	Min: 1 Max: 17 Mean: 9.17 (4.24)	N/A	Animals	60 sec.	Mean, SD, min, max	N/A	Elderly facebook users and non-users
İnal (2019)	Turkish/ Thesis/ Turkey	No	59 (33/26)	18–40	Mean: 16.10	N/A	Animals, Vegetables and Fruits, Supermarket Items	60 sec.	Mean, SD	N/A	Musicians and non-musicians
Nielsen and Waldemar (2016)	English/ Journal Article/ Denmark	Yes	41 (10/31)	≥50	Min: 0 Mean: 5	N/A	Animals, Supermarket Items	60 sec.	Mean, SD	N/A	Literate vs illiterate immigrants in Denmark	Cognition
Bozdemir (2008)	Turkish/ Thesis/ Turkey	Yes	181 (87/94)	18–80	Min: 5 Max: 13+	Middle-income class	Animals	60 sec.	Mean, SD	N/A	Pyramids and Palm Trees Test, semantic verbal fluency
Talas (2009)	Turkish/ Thesis/ Turkey	Yes	88 (44/44)	18–26 (1.52)	N/A	N/A	Animals, Vegetables and Fruits, Supermarket Items	60 sec.	Mean, SD	N/A	Right-handed, left-handed
Balcı (2016)	Turkish/ Thesis/ Turkey	No	45 (6/39)	65–83 (4.62)	Min: 5 Max: 13+	Occupation	Animals	60 sec.	Mean, SD	N/A	Dual task examination: cognitive tests balance and walking training
Kapu (2019)	Turkish/ Thesis/ Turkey	No	60 (28/32)	18–58	Min: 8 Max: 18	Income bracket	Animals	60 sec.	Mean, SD	N/A	First-degree healthy relatives of schizophrenic patients vs healthy controls	Familial risk
Çukurova (2020)	Turkish/ Thesis/ Turkey	No	82 (50/32)	18–64	Mean: 9.85	Occupation	Animals	60 sec.	Mean, SD, quarters (each 15 sec.)	Perseverations category violations	Siblings of substance-induced psychotic disorder
Aydın et al. (2017)	English/ Journal Article/ Turkey	No	88 (88/–)	25–55	Min: 5 Mean: 9.01	Income levels	Animals	60 sec.	Mean, SD	N/A	Relatives of schizophrenia patients
Berberoğlu (2018)	Turkish/ Thesis/ Turkey	No	68 (30/38)	16–30	Min: 11 Mean: 14.39	Occupation	Animals	60 sec.	Mean, SD	N/A	Siblings of schizophrenic patients
Noyan (2011)	Turkish/ Thesis/ Turkey	No	62 (33/29)	15–35	Mean: 13.94	Occupation	Animals	60 sec.	Mean, SD	N/A	First-degree relatives of patients with schizophrenia
Gürses (2009)	Turkish/ Thesis/ Turkey	No	92 (46/46)	≥18 (11.00)	Mean: 12.30 (3.11)	N/A	Animals, First Names	60 sec.	Mean, SD	N/A	Relatives of schizophrenic patients
Evlice (2016)	Turkish/ Journal Article/ Turkey	No	100 (40/60)	18–77 (13.33)	Min: 5 Max:15 (3.54)	N/A	Animals	60 sec.	Mean, SD, min, max	N/A	Neuropsychological tests with demographic comparisons	Other
Nielsen et al. (2012)	English/ Journal Article/ Denmark	No	109 (50/59)	50–87	Min: 0 Max: 16	N/A	Supermarket Items	60 sec.	Mean, SD, min, max	N/A	Turkish immigrants vs Danish elderly
Albayrak (2015)	Turkish/ Thesis/ Turkey	No	124 (20/104)	20–49	Min: 11 Max: 17	Occupation and Income	Animals	60 sec.	Mean, SD, min, max	N/A	Cognitive functions with and without procrastination
Güneş et al. (2022)	Turkish/ Journal Article/ Turkey	Yes	50 (23/27)	16–25	Min: 13 Max: 18	N/A	Supermarket Items	60 sec.	Mean, SD	Category violations	Low sleepers, Normal sleepers, Excessive sleepers
Sahin (2023)	Turkish/ Thesis/ Turkey	No	200 (113/87)	18–22	Min: 13 Max: 19	N/A	Animals	60 sec.	Mean, SD	N/A	Video gamers, Athletes, Musicians, Do not engage in any activity

Table 5. List of studies comparing people with a mental health disorder to a healthy control group with full extracted data.

General Information			Participants						Semantic Verbal Fluency			Analysis
Study ID	Language/ Type/ Country	Is SVF focus?	Clinical case	Case N (M/F)	Healthy N (M/F)	Age Min-Max (SD)	Education Min-Max (SD)	Economic status	Categories		Duration	Word count metrics		Other metrics
Balcılar (2019)	Turkish/Thesis/Turkey	No	Bipolar disorder	37 (16/21)	37 (16/21)	18–64	Mean: 12.20	N/A	Animals		60 sec.	Mean, SD		N/A
Bora et al. (2019)	English/Journal Article/Turkey	No	Bipolar disorder	71 (30/41)	50 (23/27)	15–30	Mean: 13.37	N/A	Animals		Unknown	Mean, SD		Latent class analysis
Kiylioglu (2018)	Turkish/Thesis/Turkey	No	Bipolar disorder	46 (10/36)	43 (5/38)	18–55	Min: 5 Max: 11+	Occupation and Income	Animals		60 sec.	Mean, SD		N/A
Er (2015)	Turkish/Thesis/Turkey	No	Bipolar disorder	82 (31/51)	55 (21/34)	18–65	Mean: 11.90	N/A	Animals		60 sec.	Mean, SD		N/A
Esen (2010)	Turkish/Thesis/Turkey	No	Bipolar disorder	55 (21/34)	26 (8/18)	18–64	Mean: 12.19	N/A	Animals		60 sec.	Mean, SD		N/A
Bora et al.(2007)	English/Journal Article/Turkey	No	Euthymic bipolar disorder	65 (30/35)	30 (14/16)	18–55	Mean: 12.02	N/A	Animals		60 sec.	Mean, SD		N/A
Uslu (2018)	Turkish/Thesis/Turkey	No	Bipolar disorder, Major depressive disorder	24 (12/12)	12 (6/6)	≥18	Mean: 11.60	N/A	Domestic Animals		60 sec.	Mean, SD		N/A
Arat (2015)	Turkish/Thesis/Turkey	No	Comorbid bipolar disorder, Attention deficit hyperactivity disorder	101 (42/59)	51 (22/29)	18–64	Mean: 12.90	Occupation	Animals		60 sec.	Mean, SD		N/A
Yilmaz (2014)	Turkish/Thesis/Turkey	No	Attention deficit hyperactivity disorder	30 (16/14)	30 (15/15)	16–65	Min: 8 Max: 22	N/A	Animals		60 sec.	Mean, SD, min, max		N/A
Senol (2017)	Turkish/Thesis/Turkey	No	Borderline personality disorder	21 (-/21)	20 (-/20)	18–45	Min: 8 Max: 20	Occupation	Animals, First Names		60 sec.	Median, min, max		N/A
Dinc (2014)	Turkish/Thesis/Turkey	No	Early onset depression, Late onset depression	30 (11/19)	11 (5/6)	≥60	Mean: 9.50	N/A	Animals, First Names		60 sec.	Mean, SD		N/A
Isikli (2011)	Turkish/Thesis/Turkey	No	Major depressive disorder	18 (6/12)	18 (9/9)	29–58	Min: 5 Max: 18	N/A	Animals		60 sec.	Mean, SD		N/A
Kartal (2015)	Turkish/Thesis/Turkey	No	Obsessive compulsive disorder	30 (15/15)	30 (15/15)	16–64	Min: 5 Max: 22	Occupation	Animals		60 sec.	Mean, SD		N/A
Begenen (2020)	Turkish/Thesis/Turkey	No	Obsessive compulsive disorder	79 (30/49)	41 (19/22)	18–53	Min: 5 Max: 17	Occupation	Animals		60 sec.	Mean, SD, min, max		N/A
Tukel et al.(2012)	English/Journal Article/Turkey	No	Obsessive compulsive disorder	100 (38/62)	110 (43/67)	17–50	Mean: 11.55	N/A	Animals		60 sec.	Mean, SD		N/A
Onder Uzgan et al. (2021)	English/Journal Article/Turkey	No	Obsessive compulsive disorder	66 (23/43)	75 (29/46)	18–64	Mean: 13.37	Occupation	Animals		60 sec.	Mean, SD		Perseveration, Clusters and switches (Troyer taxonomy)
Özcan et al. (2016)	English/Journal Article/Turkey	No	Obsessive compulsive disorder	51 (17/34)	21 (12/9)	18–64 (10.80)	Mean: 10.40 (.4.00)	Occupation		Animals, First Names	60 sec.		Mean, SD		N/A
Kurt (2013)	Turkish/Thesis/Turkey	No	Obsessive compulsive disorder, Panic disorder	32 (5/27)	26 (5/21)	18–60	Min: 5 Max: 11+	Occupation		Animals	60 sec.		Mean, SD		N/A
Ekinci (2017)	Turkish/Thesis/Turkey	No	Obsessive compulsive disorder, Panic disorder, Generalized anxiety disorder	45 (25/20)	29 (9/20)	18–45	Min: 5 Max: 11+	Occupation		Animals	60 sec.		Mean, SD		N/A
Kilic (2015)	Turkish/Thesis/Turkey	No	Panic disorder	31 (11/20)	31 (11/20)	18–64 (8.60)	Mean: 12.55 (3.10)	N/A		Animals	60 sec.		Mean, SD, min, max		N/A
Eren-Kocak et al. (2009)	English/Journal Article/Turkey	No	Post-traumatic stress disorder	31	20	18–64	Mean: 11.90	N/A		Animals, First Names	60 sec.		Mean, SD		N/A
Ozak (2019)	Turkish/Thesis/Turkey	No	Schizophrenia	130 (72/58)	40 (23/17)	18–55	Min: 5 Max: 11+	Occupation		Animals	60 sec.		Mean, SD, median, min, max		N/A
Aksoy-Poyraz et al. (2011)	English/Journal Article/Turkey	No	Schizophrenia	162 (99/63)	51 (32/19)	18–60	Mean: 8.90	N/A		Animals	60 sec.		Mean, SD		N/A
Mutlu et al. (2021)	English/Journal Article/Turkey	No	Schizophrenia	46 (28/18)	35 (19/16	18–64	Mean: 11.75	Occupation		Animals, First Names	60 sec.		Mean, SD		N/A
Özçelik-Eroğlu et al. (2014)	English/Journal Article/Turkey	No	Schizophrenia	16 (10/6)	8 (5/3)	Mean: 34.14 (11.00)	Mean: 11.16	N/A		Animals, First Names	Unknown		Mean, SD		N/A
Ozcelik-Eroglu et al. (2020)	English/Journal Article/Turkey	No	Schizophrenia	16 (10/6)	8 (5/3)	Mean: 34.14 (11.00)	Mean: 11.16	N/A		Animals, First Names	Unknown		Mean, SD		N/A
Sevik et al. (2011)	English/Journal Article/Turkey	No	Schizophrenia	50 (25/25)	25 (11/14)	Mean: 31.80	Mean: 11.00	Occupation		Animals, First Names	Unknown		Mean, SD		N/A
Sumiyoshi et al. (2014)	English/Journal Article/Turkey, Japan, USA	Yes	Schizophrenia	101 (68/33)	50 (34/16)	Mean: 30.83	Mean: 13.21	N/A		Animals	60 sec.		Mean, SD		N/A
Yazihan and Yetkin (2020)	English/Journal Article/Turkey	No	Schizophrenia	25	27	Mean: 22.68	Mean: 11.90	N/A		Animals	60 sec.		Mean, SD		N/A
Zagli (2011)	Turkish/Thesis/Turkey	No	Schizophrenia, Psychotic-like experience	51 (20/31)	23 (14/19)	15–65 (Mean: 41.88)	Mean: 9.38	Occupation		Animals	60 sec.		Mean, SD		N/A
Canli (2011)	Turkish/Thesis/Turkey	No	Schizotypy	22 (12/10)	22 (14/8)	15–18 (0.90)	Min: 8	Occupation		Animals, First Names	60 sec.		Mean, SD		N/A
Gürses (2022)	Turkish/Thesis/Turkey	No	Schizophrenia, Psychotic disorder	40 (23/17)	61 (25/36)	18–60	Min: 5 Max: 19	Occupation and Income bracket		Animals	Unknown		Mean, SD		N/A
Özden (2022)	Turkish/Thesis/Turkey	No	Major depressive disorder	30 (10/20)	30 (10/20)	Mean: 29.80 (7.80)	Mean: 15.45	Occupation		Animals, First Names	60 sec.		Mean, SD		N/A

Table 6. List of studies comparing people with neurodegenerative disorders to healthy controls with full extracted data.

General Information			Participants							Semantic Verbal Fluency		Analysis
Study	Language/ Publication type/ Country	Is SVF focus?	Clinical case	Case N (M/F)	Healthy N (M/F)	Age Min-Max (SD)		Education Min-Max (SD)	Economic status	Categories	Duration	Word count metrics	Other metrics
Hanagasi et al. (2002)	English/Journal Article/Turkey	No	Amyotrophic lateral sclerosis	20 (15/5)	13 (10/3)	Mean: 9.20		N/A	N/A	Animals	Unknown	Mean, SD	N/A
Buyukgok (2017)	Turkish/Thesis/Turkey	No	Apathetic and Non-apathetic Alzheimer’s type dementia	20 (8/12)	9 (5/4)	≥64 (Mean: 71.26)		Mean: 8.84	N/A	Animals	60 sec.	Mean, SD	N/A
Çabuk et al. (2020)	English/Journal Article/Turkey	Yes	Mild cognitive impairment	20 (6/14)	20 (6/14)	48–83		Min: 5 Max: 11+	N/A	Animals	60 sec.	Mean, SD, min, max	Perseveration, Clusters and switches, Category violation (Troyer taxonomy)
Dinç (2019)	Turkish/Thesis/Turkey	Yes	Mild cognitive impairment, Alzheimer’s disease	125 (58/67)	134 (53/81)	60–91 (Mean: 72.6)		Mean: 10.83	N/A	Animals	60 sec.	Mean, SD	N/A
Kalafatoglu (2015)	Turkish/Thesis/Turkey	Yes	Alzheimer’s disease	20	20	69–85 (3.31)		Min: 0 Max: 5	N/A	Animals	60 sec.	Mean, SD, min, max	Perseveration, Clusters and switches, Category violation (Troyer taxonomy)
Karaca (2015)	Turkish/Thesis/Turkey	Yes	Frontotemporal lobar degeneration, Semantic dementia	21	21	48–71 (6.19)		N/A	N/A	Animals	60 sec.	Mean, SD, min, max	Clusters and switches (Troyer taxonomy)
Kayserili (2010)	Turkish/Thesis/Turkey	No	Mild cognitive impairment, Alzheimer’s disease	61 (24/37)	106 (45/61)	≥50 (9.50)		Min: 0 Max: 11+	Occupation	Animals	60 sec.	Mean, SD	N/A
Kurt (2008)	Turkish/Thesis/Turkey	No	Mild cognitive impairment, Subjective memory complaint	28 -(9/19)	16 (5/11)	37–77 (Mean: 58.65)		Mean: 10.05	N/A	Animals	60 sec.	Mean, SD	N/A
Demirci (2018)	Turkish/Thesis/Turkey	No	Duchenne muscular dystrophy	58 (-/58)	36 (-/36)	26–67 (Mean: 40.80)		Mean: 8.10	N/A	Animals	60 sec.	Mean, SD, median	N/A
Kirac et al. (2014)	English/Journal Article/Turkey	No	Early multiple sclerosis, Clinically isolated syndromes, Definite diagnosis of multiple sclerosis	46 (16/30)	40 (14/26)	18–50 (Mean: 32.55)		Mean: 11.20	N/A	Animals	60 sec.	Mean, SD, min, max	N/A
Dogan (2020)	Turkish/Thesis/Turkey	No	Essential tremor	40 (20/20)	40 (20/20)	18–60		Min: 5 Max: 11+	N/A	Animals	60 sec.	Mean, SD, median, min, max	N/A
Sen (2014)	Turkish/Thesis/Turkey	No	Essential tremor	11 (7/4)	11 (4/7)		18–55	Min: 11 Max: 14	N/A	Animals	60 sec.	Mean, SD	N/A
Ozkul et al.(2020)	English/Journal Article/Turkey	No	Multiple sclerosis	112 (15/97)	25 (5/20)		18–65	Mean: 14.30	N/A	Vegetables and Fruits	60 sec.	Median	N/A
Ozkul et al. (2021)	English/Journal Article/Turkey	No	Multiple sclerosis	60 (16/44)	22 (6/16)		18–64 (Mean: 33.50)	Min: 12 Max: 16	N/A	Vegetables and fruits	30 sec.	Median	N/A
Tuncer et al. (2012)	English/Journal Article/Turkey	No	Multiple sclerosis	25 (-/25)	17 (-/17)		20–50 (Mean: 34.33)	Mean: 10.38	N/A	Animals	60 sec.	Mean, SD	N/A
Diker (2014)	Turkish/Thesis/Turkey	No	Multiple sclerosis (Clinically isolated syndrome)	30 (10/20)	20 (7/13)		19–51 (Mean: 31.82)	Min: 5 Max: 14 Mean: 12.60	N/A	Animals	90 sec.	Mean, SD, min, max	N/A
Baran (2008)	English/Thesis/Turkey	No	Non-demented Parkinson’s disease	18 (15/3)	18 (12/6)		Mean: 64.52	Mean: 10.30	N/A	Animals	60 sec.	Mean, SD	N/A
Degirmenci (2016)	Turkish/Thesis/Turkey	No	Parkinson’s disease	32 (18/14)	32 (12/20)		≥50 (Mean: 59.90)	Min: 5 Max: 14	N/A	Animals	60 sec.	Mean, SD, min, max	N/A
Demirci (2010)	Turkish/Thesis/Turkey	No	Parkinson’s disease	27 (17/10)	12 (6/6)		Mean: 45.90	Mean: 9.80	N/A	Animals	60 sec.	Mean, SD	N/A
Yildirim (2012)	Turkish/Thesis/Turkey	No	Parkinson’s disease	39 (31/8)	39		48–81	10.65 (4.36)	N/A	Animals	60 sec.	Mean, SD	N/A
Yılmaz et al. (2020)	English/Journal Article/Turkey	No	Parkinson’s disease	55 (40/15)	20 (9/11)		Mean: 65.27	Mean: 6.95	N/A	Animals	Unknown	Mean, SD	N/A
Öz et al.(2022)	Turkish/Journal Article/Turkey	Yes	Mild cognitive impairment, Alzheimer’s disease	144 (62/82)	72 (26/46)		Mean: 73.68	Mean: 10.49	N/A	Animals	60 sec.	Mean, SD	N/A
Altun (2022)	Turkish/Thesis/Turkey	Yes	Multiple sclerosis	20 (4/16)	20 (4/16)		26–56 (Mean: 39.90)	Min: 5 Max: 16	N/A	Animals	60 sec.	Mean, SD, median, min, max	Clusters and switches (Troyer taxonomy)
Demiray and Ertan (2023)	Turkish/Journal Article/Turkey	No	Parkinson’s disease, Essential tremor	60 (37/23)	13 (7/6)		40–80 (Mean: 60.78)	Mean: 9.02	N/A	Animals	Unknown	Mean, SD, min, max	Perseveration

Table 7. List of other studies comparing a group of Turkish speakers with a disease or disorder to healthy controls with full extracted data.

General Information			Participants							Semantic Verbal Fluency		Analysis
Study ID	Language/ Publication type/ Country	Is SVF focus?	Clinical case	Case N (M/F)	Healthy N (M/F)	Age Min-Max (SD)	Education Min-Max (SD)	Economic status		Categories	Duration	Word count metrics	Other metrics	Disease type
Çakar (2020)	Turkish/Thesis/Turkey	No	Epilepsy	64 (21/43)	46 (16/30)	18–64 (Mean: 31.28)	5–16 (Mean: 12.68)	N/A		Animals	60 sec.	Mean, SD	N/A	Epilepsy
Evlice et al. (2016)	Turkish/Journal Article/Turkey	No	Epilepsy	30 (15/15)	21 (13/8)	18–64 (Mean: 33.19)	Min: 5	N/A		Animals	60 sec.	Mean, SD, min, max	N/A
Kizil (2015)	Turkish/Thesis/Turkey	No	Epilepsy	30 (13/17)	30 (13/17)	18–46 (5.93)	Min: 5 Max: 14	N/A		Animals	60 sec.	Mean, SD	N/A
Sahin (2022)	Turkish/Thesis/Turkey	No	Epilepsy	30 (11/19)	15 (5/10)	18–45 (Mean: 27.38)	Mean: 13.96	N/A		Animals	60 sec.	Mean, SD, median, min, max	Perseveration
Ozer Celik et al. (2015)	Turkish/Journal Article/Turkey	No	Epilepsy, Psychogenic nonepileptic seizures	31 (9/22)	20 (5/15)	Mean: 29.50	Mean: 10.60	Occupation		Animals	Unknown	Mean, SD	N/A
Balcik (2019)	Turkish/Thesis/Turkey	No	Juvenile myoclonic epilepsy	60 (25/35)	30 (13/17)	15–55 (8.50)	Min: 5 Max: 14	N/A		Animals	60 sec.	Mean, SD	N/A
Cevik (2011)	Turkish/Thesis/Turkey	No	Juvenile myoclonic epilepsy	20 (6/14)	16 (5/11)	16–40 (6.89)	5–15 (3.26)	N/A		Animals	60 sec.	Mean, SD	N/A
Sezikli et al. (2018)	English/Journal Article/Turkey	No	Juvenile myoclonic epilepsy	45 (11/34)	15 (4/11)	Mean: 23.20	Min: 5 Max: 14	N/A		Animals	Unknown	Mean, SD	Perseveration
Kilic (2012)	Turkish/Thesis/Turkey	No	Mesial temporal lobe epilepsy	28 (16/12)	36 (15/21)	18–55 (Mean: 33.80)	Mean: 9.50 (4.00)	Occupation and Income		Animals	60 sec.	Mean, SD	N/A
Uslu (2015)	Turkish/Thesis/Turkey	No	Mesial temporal lobe epilepsy	79 (30/49)	30 (16/14)	18–57 (9.22)	Min :5 Max: 14	N/A		Animals	60 sec.	Mean, SD, median,min, max	N/A
Cengiz (2018)	Turkish/Thesis/Turkey	No	Abnormal cranial imaging patients	30 (14/16)	30 (14/16)	18–50 (Mean: 36.95)	5–16 (4.70)	N/A		Animals	60 sec.	Mean, SD, median	N/A	Neurological
Alibas et al.(2017)	English/Journal Article/Turkey	No	Acromegaly	42 (19/23)	44 (19/25)	18–64 (Mean: 41.30)	Mean: 8.08	N/A		Animals	60 sec.	Mean, SD	N/A
Tutan (2019)	Turkish/Thesis/Turkey	No	Chronic subjective tinnitus	60 (38/22)	30 (18/12)	21–60	5–16 (4.90)	N/A		Animals	60 sec.	Mean, SD, median, min, max	N/A
Bakar and Bakar (2010)	English/Journal Article/Turkey	No	Hydrocephalus	15 (9/6)	15 (9/6)	18–64	N/A	N/A		Animals	60 sec.	Mean, SD	N/A
Ulasoglu (2010)	Turkish/Thesis/Turkey	No	Korsakoff amnestic syndrome	13 (10/3)	13 (10/3)	18–64 (Mean: 39.30)	Mean: 10.54	N/A		Animals	60 sec.	Mean, SD	N/A
Buyukgok et al. (2021)	English/Journal Article/Turkey	No	Normal pressure hydrocephalus	30 (13/17)	30 (12/18)	≥60 Mean: 67.72 (±8.13)	Mean: 9.01 (3.90)	N/A		Animals	60 sec.	Median, interquartile	N/A
Eray (2013)	Turkish/Thesis/Turkey	No	Restless legs syndrome	15 (5/10)	15 (5/10)	≥18 Mean: 47.95	Mean: 8.8 (4.10)	N/A	Animals		60 sec.	Mean, min, max	N/A	Neurological
Hatipoglu et al. (2022)	English/Journal Article/Turkey	No	Acromegaly	33 (10/23)	30 (8/22)	Mean: 46.60	Min: 5 Max: 16	N/A	Animals		60 sec.	Median, interquartile	N/A
Ersan (2014)	Turkish/Thesis/Turkey	No	Alcohol addiction	29 (9/10)	30 (9/11)	18–64 Mean: 45.80	Mean: 12.55	Occupation	Animals		60 sec.	Mean, SD	Perseveration	Addiction
Kocuk (2010)	Turkish/Thesis/Turkey	No	Alcohol addiction	18 (18/-)	18 (18/-)	Mean: 47.69	Mean: 11.33	Occupation	Animals		60 sec.	Mean, SD	N/A
Demir and Uluğ (2002)	Turkish/Journal Article/Turkey	No	Alcohol addiction	34 (34/-)	11 (11/-)	Mean: 42.78	Mean: 10.00	N/A	Animals, First Names		Unknown	Mean, SD	N/A
Çakmak (2022)	Turkish/Thesis/Turkey	No	Opioid use disorder	93 (93/-)	70 (70/-)	19–48 Mean: 27.75	8–18 (Mean: 11.00)	Income Levels	Animals		60 sec.	Mean, min, max	N/A
Sen Ozdemir (2020)	Turkish/Thesis/Turkey	No	Neuromyelitis optica spectrum	20 (3/17)	23 (4/19)	23–54 Mean: 38.05	Min:8 Max:16	N/A	Animals		60 sec.	Mean, SD, median, min, max	N/A	Autoimmune
Bilgin Topcuoglu et al. (2018)	English/Journal Article/Turkey	No	Sarcoidosis	21 (7/14)	21 (9/12)	18–64 Mean: 43.80	Mean: 7.59	N/A	Animals		60 sec.	Mean, SD	N/A	Inflammation
Altunkaynak et al. (2019)	English/Journal Article/Turkey	No	Behcet’s Disease	30 (20/10)	20 (13/7)	Mean: 33.07	Mean: 11.74	N/A	Animals, First Names		Unknown	Mean, SD	N/A
Kandemir (2006)	Turkish/Thesis/Turkey	No	Infratentorial strokes	19 (17/2)	19 (5/14)	40–60 Mean: 48.37	5–15 (Mean: 7.30)	N/A	Animals		60 sec.	Mean, SD, quarters (each 15 sec.)	N/A	Stroke
Erdal et al.(2021)	English/Journal Article/Turkey	No	Isolated cerebellar infarctions	23 (17/6)	22 (15/7)	18–64 Mean: 53.56	Min: 8	N/A	Animals		60 sec.	Mean, SD, median	N/A
Temel (2020)	Turkish/Thesis/Turkey	No	Thalamic hemorrhage	28 (18/10)	28 (18/10)	42–80 Mean: 60.82	5–16 (3.80)	N/A	Animals		60 sec.	Mean, SD	N/A
Sogutlu and Alaca (2019)	Turkish/Journal Article/Turkey	No	Memory complaints	56 (15/41)	55 (20/35)	≤55 Mean: 36.45	Min: 5 Max: 14	N/A	Animals		Unknown	Mean, SD	N/A	Other
Akdemir (2021)	Turkish/Thesis/Turkey	No	Chronic pain	91 (28/63)	70 (22/48)	18–64 Mean: 41.33	Mean: 10.66	Occupation	Animals		60 sec.	Mean, SD	N/A
Töret and Özdemir (2021)	English and Turkish/Journal Article/Turkey	No	Blindness	20 (14/6)	20 (14/6)	≥18	Min: 11 Max: 16	N/A	Animals		60 sec.	Mean, SD	N/A
Hangun (2022)	Turkish/Thesis/Turkey	No	COVID-19	40 (17/23)	40 (17/23)	18–75 Mean: 33.36	Min: 5 Max: 22	N/A	Animals		60 sec.	Mean, SD	N/A
Iscen (2022)	Turkish/Thesis/Turkey	No	Glucocerebrosidase mutation carriers	16 (7/9)	18 (8/10)	Mean: 17.47	Min: 30 Max: 60	N/A	Animals		60 sec.	Mean, SD	N/A
Cengiz-Al et al. (2023)	English/Journal Article/Turkey	No	Cavum septum pellucidum	26 (9/17)	30 (14/16)	18–50 Mean: 36.95	Min: 5 Max: 16 Mean: 10 (4.70)	N/A	Animals		60 sec.	Mean, SD	N/A
Arisoy et al. (2023)	English/Journal Article/Turkey	No	Obstructive sleep apnea syndrome	78 (70/8)	26 (13/13)	18–50 Mean: 38.57	Mean: 11.97	N/A	Animals		60 sec.	Mean, SD	N/A

Quality assessment

As we were interested in the SVF results, which may not have been the main outcome measure of the studies included, we did not conduct a formal bias assessment of the studies themselves. Rather, we noted the level of detail with which the data collection and analysis procedures for SVF scores was reported, and summarized our findings in an overall assessment of the studies. Studies received one point for each relevant detail (e.g., whether words were audio recorded or recorded in writing; whether a reference was given that describes the exact scoring method; etc.). For each study, two members of the review team independently performed this quality assessment and disagreements were resolved by discussion between three members of the review team (RYK, MKW, SM). Data synthesis was independent of the quality assessment.

Results

RQ1 and RQ2: Categories and scoring metrics

In general, animal naming is the preferred category in SVF, because it is least affected by country, cultural, educational and generational differences (Ardila et al., 2006). As in other languages, for Turkish, animals are the most frequently used category (n = 114, 94%). Other categories were first names (n = 14, 12%), vegetables and fruits (n = 6, 5%) and supermarket items (n = 7, 6%). Even more rarely used categories included vehicles, clothes, parts of the body, furniture, famous people, breakfast items, food, beverages, and household items. Most of the rarer categories appeared in normative studies, which tended to provide verbal fluency data for a range of possible categories.

The administration time for SVF varied. Sixty seconds was used in 87% of studies (n = 105), 12% of studies did not indicate the duration of the test (n = 14), 2% used 90 seconds (Diker et al., 2016; Özdemir, 2015) (n = 2), and one (1%) study with a dual task design used 30 seconds (Özkul et al., 2021).

In terms of the metrics used, all studies reported the total number of words generated. Only four studies provided the number of words provided in each quarter of the 60 second period (Çukurova, 2020; Kandemir, 2006; Özdemir, 2015; Özdemir & Tunçer, 2021). In terms of more in-depth analyses, 12 studies (Aki et al., 2022; Çabuk et al., 2020; Çukurova, 2020; Demiray & Ertan, 2023; Ersan, 2014; İlkmen & Büyükişcan, 2022; Kalafatoglu, 2015; Sahin, 2022; Sezikli et al., 2018; Tumaç, 1997; Uzgan et al., 2021; Yazici, 2019) (10%) reported perseverations, and six studies (Çabuk et al., 2020; Çukurova, 2020; Güneş et al., 2022; İlkmen & Büyükişcan, 2022; Kalafatoglu, 2015; Yazici, 2019) (5%) reported category violations. Clustering and switching was also used in only five studies (Altun, 2022; Çabuk et al., 2020; Kalafatoglu, 2015; Karaca, 2015; Uzgan et al., 2021) (4%). None of these papers provided a full translation for the original English taxonomy, although three papers listed the groups of animals used. Çabuk et al. (2020) and Kalafatoglu (2015) created a limited number of animal groups, namely farm, pets, forest, and zoological categories (insects, birds, and fish). Both studies also report perseverations. In their comparative study of frontotemporal lobe degeneration and semantic dementia, Karaca (2015) used different groups, namely poultry, ovine livestock, bovine livestock, forest, farm, birds, and insects. All studies were based on manual analysis, and differences between groups were established using standard inferential statistics. Bora et al. (2019) used manually established SVF scores as parameters for latent class analysis to explore group differences.

In the following subsections, we present an overview of the main tasks and metrics for non-normative studies. These fall into three categories: studies of bilingualism; comparisons between groups of healthy native speakers; and comparisons between patients and healthy control groups. Since a full meta analysis of Turkish SVF data is beyond the scope of this review, and the number of studies across all categories is substantial, we only summarize findings for studies where SVF performance was a main focus.

Bilingualism

We found only one study that focused on speaking another language in addition to Turkish, which was a Turkish Masters dissertation (c.f. Table 4). Yazici (2019) compared the executive functions of bilingual Kurdish-Turkish speakers from Eastern and Southeastern Anatolia (n = 80) to that of monolingual Turkish speakers from across the country (n = 80). Animal naming was used, and word count, perseverations, and category violations were reported. Participants had 9 years or more of education and the age range was 20–54 years. Bilingual individuals’ language proficiency levels were not formally determined, and there was no information about proficiency in other languages. While the study did not focus on SVF performance, the author argued that cultural aspects, such as languages spoken, need to be accounted for in normative studies of executive function in Turkish.

Comparison between groups of healthy native speakers

Excluding normative studies, we found 21 studies conducted with healthy individuals, which are summarized in Table 4. Only two studies were conducted outside of Turkey. These were studies of minorities in Denmark (Nielsen & Waldemar, 2016; Nielsen et al., 2012) carried out with the help of an interpreter, and did not focus on bilingual or multilingual skills. Five studies focused on people aged 50 years and over. The topics of these studies fell into five distinct categories:

Physical activity

Exercise may have a positive effect on cognitive abilities due to the acceleration of blood circulation and increase of oxygen delivery to the brain (Mandolesi et al., 2018). Three studies examined the overall effect of exercise on cognition and used SVF as part of a standard assessment battery (Gökçe et al., 2021): tennis (Gökçe, 2020); fencing or swimming (Yeniçeri, 2019); and balance training.

Cognitive reserve

Lifetime experiences, such as educational and occupational attainment, literacy attainment and the involvement in cognitive and socially stimulating activities, are thought to increase the efficacy of cognitive processing in older age (for a review, see Arenaza-Urquijo et al. (2015)). Ekin and Çebi (2021) examined the concept of cognitive reserve and emotion regulation in older people, while Yıldırım and Ogel-Balaban (2021) studied older people’s use of social media (e.g., Facebook). İnal (2019) compared professional musicians with non-musicians to investigate the effects of active interest in music. In all three studies, SVF was used as part of a standard assessment. Nielsen and Waldemar (2016) studied the relationship between SVF and literacy of healthy older Turkish immigrants in Denmark. Two categories were used, animals and supermarket items. Illiterate immigrants performed worse than literate ones on animal naming, but equally well on the supermarket category.

Cognition

In their study of the effect of various forms of training, including dual task training, on reducing fall risk, Balcı (2016) used SVF as part of their assessment battery. Despite the title of the thesis, Bozdemir (2008) focused on establishing Turkish normative data for the Pyramid and Palm Trees test, a semantic dementia test developed by Howard and Patterson (1992). SVF was administered as part of a larger battery of tests for comparison purposes. In the only study to specifically highlight SVF performance, Talas (2009) studied the relationship between magical ideation, handedness, and verbal fluency. They found a negative correlation between magical ideation scores and SVF scores, where the higher the magical thinking, the poorer the SVF performance.

Familial risk

Studies in this group include healthy individuals who are first-degree relatives of people with a mental health condition and are therefore considered to have familial risk factors compared to the general population. The conditions included are schizophrenia (Aydın et al. 2017; Berberoğlu, 2018; Gürses, 2009; Kapu, 2019; Noyan, 2011) and substance-induced psychotic disorder (Çukurova, 2020). Bora et al. (2019) used Latent Class Analysis (LCA) to classify the neurocognitive performance of euthymic children of parents with bipolar disorder. They found that the performance of the children of people with bipolar disorder could be categorized as: severe impairment; intermediate impairment; and good performance. SVF word count was significantly lower for children classified in the severe and intermediate impairment groups, compared to healthy controls and the “good performance” group.

Other

Studies in this category do not share a common framework in terms of their focus. Albayrak (2015) investigates procrastination in young adults, using SVF as part of their assessment battery. Evlice (2016) reports the effects of age, gender, and education on healthy participants’ performance on a standard battery of neuropsychological tests. SVF performance is negatively correlated with age (where the older the age, the poorer the SVF performance) and positively correlated with education (where the higher the level of education, the higher the SVF performance). In their comparison of Turkish immigrants and Danish citizens, Nielsen et al. (2012) examined the cross-cultural applicability of SVF, but without a monolingual Turkish control group. They used supermarket items as the test category. Turkish participants were administered the tests in Turkish using an interpreter. Nielsen et al. found that Turkish immigrants with higher acculturation produced more items, while older immigrants produced fewer. Güneş et al. (2022) investigated SVF performance in medical school students in terms of low, normal and excessive sleep duration. They revealed that sleep duration has no significant influence on SVF performance. In another study, Sahin (2023) compared video gamers, athletes, musicians and participants who do not engage in any activity and found no significant difference between groups.

Comparison between patients and healthy control groups

Ninety-two out of 121 studies (76%) compared the performance of people with an impairment or condition to a healthy control group.

Mental health

Thirty-three of 92 studies (36%), summarized in Table 5, involved mental health conditions. In descending order of frequency, conditions included schizophrenia, bipolar disorder, obsessive compulsive disorder, depression, hyperactivity, borderline personality disorder, panic disorder, post-traumatic stress disorder, and schizotypy.

Only Sumiyoshi et al. (2014) examined verbal fluency as a main focus of their study. Both verbal fluency types (semantic fluency and letter fluency) were investigated for schizophrenic patients in three languages: Japanese, Turkish and English. Like Turkish, Japanese is an agglutinative language, which uses affixes to indicate most relevant grammatical information. For Japanese and Turkish, performance was also compared to healthy controls. In Japanese, which uses two syllable-based scripts and one logographic script, letter fluency was more impaired than semantic fluency; in Turkish, which uses an alphabetic script, semantic fluency was more impaired than letter fluency. Uzgan et al. (2021) were the only authors to examine perseverations and number of switches in addition to SVF word count. There were no differences in any of the SVF measures between people with Obsessive-Compulsive Disorder and healthy controls. Unfortunately, the study, which used animals as a category, does not provide information on how the animal groups were formed.

Neurodegenerative diseases

Twenty-four of 92 studies (26%) studied neurodegenerative diseases (c.f. Table 6). Conditions that are frequently examined are the dementias, in particular Alzheimer’s Disease (AD), Mild Cognitive Impairment (MCI), Parkinson’s Disease, Multiple Sclerosis, and Essential Tremor.

Most studies used animals as their category, except for two studies which used the vegetables and fruits category. The data collection period was 60 seconds, except for Diker (2014) (90 seconds), Demiray and Ertan (2023); Hanagasi et al. (2002); Yılmaz et al. (2020) (no information), and Özkul et al. (2021) (30 seconds). Özkul et al. (2021) also used a novel scoring method, the number of correct words per second, which is an uncommon way of reporting SVF performance.

Five studies, which all use the standard animal category, focus on SVF. The longitudinal study by Dinç (2019), which compared MCI patients, MCI patients who progressed to AD, and healthy controls, used the standard word count metric. They found that SVF was a good predictor of conversion to AD. The other three studies (Altun, 2022; Çabuk et al., 2020; Kalafatoglu, 2015; Karaca, 2015) used a clustering and switching approach, but defined their own categories. Çabuk et al. (2020) and Kalafatoglu (2015) created a limited number of animal groups, namely farm, pets, forest, and zoological categories (insects, birds, and fish). Both studies also report perseverations. In their comparative study of frontotemporal lobe degeneration and semantic dementia, Karaca (2015) used different groups, namely poultry, ovine livestock, bovine livestock, forest, farm, birds, and insects. Altun (2022) stated that the original Troyer animal groups were used, but did not provide information on how they were handled if a different animal name existed. The study excluded perseverations for the total word count, but included them for the cluster analysis. If subcategories were provided, the parent category name was excluded while creating clusters. For example, if “fish” was mentioned along with different fish names (e.g., salmon, sardine), the word “fish” was omitted.

Çabuk et al. (2020), comparing patients with amnestic MCI to healthy controls, found no difference in clustering and switching related parameters for SVF performance, but patients with amnestic MCI produced more perseverations. Kalafatoglu (2015) reported smaller SVF cluster sizes in patients with probable dementia compared to healthy controls. In Karaca (2015)’s study of patients with frontotemporal lobe degeneration, they produce fewer switches than healthy controls. Another study, Altun (2022), focused on multiple sclerosis patients and found that multiple sclerosis patients produce significantly fewer total words compared to healthy controls. No difference was observed between the groups in terms of mean cluster size, but fewer switches were produced in those with multiple sclerosis.

Other conditions

The remaining 35 (29%) studies, summarized in Table 7, covered six types of conditions, most of which are neurological. Ten studies focus on epilepsy, followed by addiction and stroke. All studies reported animal naming. Two studies added first names as a second category (Altunkaynak et al., 2019; Demir & Uluğ, 2002). In addition to word count, 3 studies report perseverations (Ersan, 2014; Sahin, 2022; Sezikli et al., 2018), while Kandemir (2006) reports performance every 15 seconds. SVF was always used as a supporting tool, never as a central focus of the research studies.

RQ3: Normative studies

We found a total of seven normative studies that focus on SVF performance. Two further studies, Bozdemir (2008) and Evlice (2016), also collected data on SVF performance across a large sample of participants, but did not specifically aim to produce normative data for SVF. The details of the studies are given in Table 3 in order of publication year.

Except for Tuncer (2012), all studies recruited a balanced number of male and female participants. While Özdemir and Tunçer (2021) focused only on people aged 60 years and older, the other six studies recruited participants across the entire adult age range. All normative studies report level of education based on the Turkish national education system. Only Tuncer (2012) additionally explored illiteracy. No socioeconomic status information is reported, except Tumaç (1997). The number of participants varied between 58 (Özdemir & Tunçer, 2021) and 1431 (İlkmen & Büyükişcan, 2022).

In terms of category type, the most comprehensive study was conducted by Tuncer (2012), who collected SVF data for six different categories, namely animals, vegetables and fruits, vehicles, clothes, parts of the body, and furniture, from 400 participants. Tuncer (2012) was also the only study that focused on literacy. Two studies (Özdemir, 2015; Özdemir & Tunçer, 2021) examined five categories in addition to animals, namely breakfast items, famous people, food, beverages, and household items. Tuncer (2012) and Özdemir (2015) also list the most frequently used words and the first words produced for each category (e.g., dog, cat, horse, and donkey, for the animal category). While İlkmen and Büyükişcan (2022); Şentürk (2019); Tumaç (1997) focused only on animal naming, Aki et al. (2022) examined both animal names and first names.

Usually, the time allowed for participants to produce the words is 60 seconds. In the only study using 90 seconds (Özdemir, 2015), the total time was divided into 3 equal parts with 30 second chunks and the differences between the chunks were examined. This is equivalent to the common method implemented by İlkmen and Büyükişcan (2022) of splitting 60 seconds into 15 second chunks and counting the number of words produced in each chunk. Word counts are reported excluding perseverations and category violations. Aki et al. (2022) and Tumaç (1997) report perseverations and İlkmen and Büyükişcan (2022) report perseverations and category violations in addition to word counts. No other analysis approaches were found.

Tumaç (1997), an unpublished Masters dissertation, reports normative results for a test battery that is sensitive to frontal lobe dysfunction. Considering age and education levels together, nine different groups were evaluated based on three age groups (young (15–18 years), middle-aged (32–45 years) and older (50–75 years)) and three education levels (low-medium-high) using animal naming. In terms of total word count, the highest score was achieved by the middle-aged group with the highest education (Mean = 26.65, SD = 5.04), while the older group with the lowest education (Mean = 19.25, SD = 4.46) and the middle age group with the lowest education (Mean = 19.40, SD = 3.66) resulted in the lowest scores. The older age group with the lowest education produced the highest number of perseverations (Mean = 1.05, SD = 0.94), followed by the middle-aged group with the highest education (Mean = 0.75, SD = 1.02). Although the participants’ occupational information was presented in the background demographic information, it was not included in the analysis.

In Tuncer (2012)’s study, SVF scores were higher than letter fluency scores. The category yielding the highest scores was body parts, followed by fruits and vegetables, animals, clothes, vehicles, and furniture. Highly educated participants had higher scores; there was no difference with people who had only a few years of education and those who were illiterate. The 25–34 years age group had the highest scores, and the 65+ age group had the lowest.

Özdemir (2015) confirms Tuncer’s results with regard to age and education in a sample of 150 participants aged between 15 and 81 years. In an in-depth study of the effect of education in older people aged 60–81, Özdemir and Tunçer (2021) showed that education level affected SVF scores, where the mean number of words every 15 seconds decreased linearly, but not chronological age.

Aki et al. (2022) report results from 415 participants. Age, gender, and level of education affected performance on SVF for first names. Higher scores on this task are linked to female gender, 9 or more years of education, and being in the 15–24 and 25–34 age groups. Animal SVF performance, in contrast, is only affected by level of education. Those with 12 or more years outperform those with 9–11 years of education, who in turn produce more words than those with only 5–8 years. There were not sufficient perseverations in either the animal or the names SVF tasks to warrant analysis. Şentürk (2019)’s study of the SVF performance of 200 healthy volunteers aged between 18 and 49 years confirms a strong effect of level of education with higher education associated with better SVF performance. In addition, male participants produced significantly more words than females.

İlkmen and Büyükişcan (2022) conducted the largest normative study in terms of number of participants with a total of 1431 healthy participants (male = 727, female = 704) from Istanbul. Eight different age groups (ranging from 18 to 89 years) were compared for four educational levels (elementary to postgraduate). They found that SVF performance was significantly negatively correlated with age but is positively related to education. Also, they emphasized that there was no difference in SVF performance between genders. Total word count decreased for people aged 50 and older.

Discussion

To the best of our knowledge, this is the first systematic review focusing on available semantic fluency data for native speakers of Turkish. We reviewed normative and non-normative studies that report data on SVF frequencies that were produced in Turkish by healthy native speakers of the Turkish language, including studies where healthy native speakers formed a control group.

We found that, while SVF is a key part of neuropsychological assessment in studies of Turkish speakers, there is a paucity of normative data for all categories typically administered. In addition, one of the categories, first names, appears specific to Turkish. Unlike categories such as animals, fruits, or supermarket items, first names do not easily lend themselves to analyses that focus on the structure of a person’s mental lexicon.

Most studies limit themselves to reporting word counts. We only found 15 (12%) studies that used more complex metrics. In cases where sequences were annotated with clusters and switches, the rules for scoring animal groups were not published, and the animal groups did not cover all of the groups defined by Troyer and collaborators (Altun, 2022; Çabuk et al., 2020; Kalafatoglu, 2015; Karaca, 2015; Uzgan et al., 2021). The lack of categorization rules may explain why studies do not consider additional clustering and switching SVF metrics in Turkish studies. Moreover, without these rules, clinicians cannot consider clustering and switching in their patients’ assessments. While word counts provide clinicians with a quick and easy measure, more detailed qualitative analysis for scoring SVF data can provide additional insights into human cognitive performance (Abwender et al., 2001; Mayr & Kliegl, 2000; Troyer, 2000; Troyer et al., 1997). Therefore, future work should provide normative data with the categorization rules for the switching and clustering metrics in Turkish.

Our review revealed that there were very little data on the performance of bilingual versus monolingual speakers in Turkish, which is a clear gap in the literature given that 6.5 million Turkish speakers live in the diaspora abroad (Turkish Ministry of Foreign Affairs, Turkish, 2022). Within Türkiye itself, several different languages are spoken, the most prominent of which is Kurdish. The Turkish Demographic and Health Survey from 2003 reports that 83% of Turkish people speak Turkish as their native language, 14% Kurdish, 2% Arabic, and 1% other languages (Koc et al., 2008). The only study in our data set that actually incorporates bilingualism involved Kurdish speakers in Turkey (Yazici, 2019). Two studies compared Turkish immigrants to Denmark to older native speakers of Danish (Nielsen & Waldemar, 2016; Nielsen et al., 2012), but they do not focus on bilingual competence.

Previous studies comparing bilinguals’ and monolinguals’ SVF performance in other languages have shown mixed results (Bialystok et al., 2008; Luo et al., 2010; Paap et al., 2017; Patra et al., 2020; Sandoval et al., 2010). While monolinguals have been found to produce significantly more correct words than bilinguals (e.g., Sandoval et al., 2010), the bilingual disadvantage is no longer found when the groups are matched for vocabulary (e.g., Bialystok et al., 2008). However, both the monolinguals and bilinguals were tested in English, which may not have been the dominant language for the bilingual group and a lack of proficiency in English may explain their disadvantage. In a recent systematic review (Giovannoli et al., 2023), more than half of the studies did not report significant differences between monolinguals and bilinguals on SVF, a small number of studies demonstrated some bilingual difficulties but no study reported a bilingual advantage. As in our own review, most studies have assessed participants using the “animal” category, with only three studies (Gollan et al., 2002; Keijzer & Schmid, 2016; Sandoval et al., 2010) adopting multiple categories. However, not all categories may be suitable for individuals living in different countries or speaking different languages; moreover, while animals may be the best cross linguistic category (Ardila, 2020), categories that draw upon everyday experience are better for assessing lower educated individuals (Nielsen & Waldemar, 2016).

When examining the provenance of the original 130 publications, we found that 82 studies (63%) were PhD or Master’s theses, and only 48 studies (37%) were peer-reviewed papers. All theses, except for Baran (2008), were written in English. Most of them were prepared by medical doctors during their specialization training in areas such as neurology or psychiatry. As noted in the Methods section, only 9 theses (Baysal Kiraç, 2012; Çabuk, 2018; Mutlu, 2018; Önder, 2019; Özcan, 2010; Özçelik-Eroğlu, 2012; Sezikli, 2014; Töret, 2019; Yavuz-Demiray, 2011) were published as peer-reviewed versions. Of the 48 peer-reviewed studies, 38 (79%) were published in English, nine (21%) in Turkish, and one (Töret & Özdemir, 2021) in both languages. Normative studies available in English were Aki et al. (2022); İlkmen and Büyükişcan (2022). This makes it very difficult for international researchers and clinicians to access data on SVF performance in Turkish.

Limitations

Relevant studies are often published in the grey literature, i.e., not by commercial publishing organizations. In this review, we focused on an easily searchable type of grey literature, namely dissertations. Other types, such as preprints only published on preprint archives, government white papers, or hospital reports (Adams et al., 2016), were not considered. There was one normative study for which we were unable to obtain the full text. Bingöl et al. (1994) is frequently cited as a key normative study, but no information about study design and results could be found. The other study, Tumaç (1997), was found through a hand search of references (Cengiz-Al et al., 2023; İlkmen & Büyükişcan, 2022) where they cited the study as a normative study. Tumaç (1997)’s study was added into this review and study details were shared in Table 3. Yet, given our difficulty accessing Bingöl et al. (1994)'s study, it is unlikely that their normative data are available to other researchers and clinicians working in Turkish.

There were also a few terminological issues that made our search more difficult. For example, we found two studies that reported SVF as a subordinate task of COWAT, which is why it was included in the original search (Çakar, 2020; Kiraç et al., 2014). Ersan (2014) used “Word List Generation” for semantic fluency, and two studies (Ersan, 2014; Özcan et al., 2016) viewed category fluency as having two parts, animal naming and alternate fruit and animal naming.

Finally, due to the heterogeneity of the studies reported, we did not perform a meta-analysis of the SVF scores reported in the literature. Instead, we hope that researchers can use the groundwork laid in this paper to perform meta-analyses for smaller, well-defined groups of conditions.

Implications

The results of our systematic review indicate that individuals administering the SVF in non-clinical Turkish populations tend to use the animal naming category most frequently. Moreover, most studies simply report the total number of words produced. While a small number of studies provide more in-depth analyses that include clustering and switching, these studies do not provide the rules for scoring the various animal groups, making it difficult for individuals to apply these metrics to their own data. Therefore, there is a need for better Turkish normative data, which include the various semantic categories administered, the various scoring metrics and the category grouping rules.

While a simple count of the number of words produced provides a quick and easy measure of SVF performance, applying Troyer’s scoring method manually is a little more time-consuming as it requires careful reanalysis of the sequence of words produced; clinicians may have limited time to spend with their patients and so using additional scoring metrics is less feasible. Moreover, the established procedure specifically addresses animal naming, but offers insufficient subgroups for many animals (local animals, mythological animals etc.). Some names may be located in more than one group (e.g., parrot can be considered within the human use category as a pet or the zoological category as a bird), which makes cluster identification difficult (Woods et al., 2016). Researchers have been investigating alternative approaches to identifying semantic clusters using generalizable, fast, and robust models that can be easily adapted to different languages and scenarios. Frequently investigated computational linguistics approaches are lexical relations (WordNet) (Paula et al., 2018; Quaranta et al., 2019), distributional semantics (word2vec, glove etc.) (Holmlund et al., 2019; Kim et al., 2019; Linz et al., 2017; Voppel et al., 2021), and transformer language models (BERT) (Alaçam et al., 2022).

Conclusions

In summary, animal naming appears to be the preferred category when SVF is administered to Turkish-speaking individuals. Moreover, those administering the test appear to opt for the quick and easy method of scoring the SVF by simply considering the number of unique words generated. However, the SVF is considered a multifactorial test and so this score may not entirely represent an individual’s performance; other scoring metrics such as cluster size and switching (Troyer, Moscovitch, Winocur, Leach, et al., 1998) may be differentially spared and impaired. Therefore, the methodology for producing semantic subcategories as well as normative data for the various SVF metrics is needed in Turkish to allow clinicians and researchers to understand their patients’ cognitive profiles qualitatively.

Author contributions

RYK, SM, and MKW designed the review. RYK performed the literature search, and RYK, KK, SM, and MKW participated in abstract screening and full text review. RYK and KK extracted data. RYK wrote the first draft of this paper, which was edited by MKW and SM. All authors read and commented on the final draft.

Prospero statement

This review was registered on Prospero with the protocol number CRD42020201585 and passed the suitability examination on 27 October 2020.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The paper presents all included studies together with the full extracted data. The tables are also provided as a supplementary Excel file. We can provide access to the Covidence repository associated with this review on request.

Additional information

Funding

This study was supported by the Ministry of National Education of Türkiye as part of Rabia Yasa Kostas’s PhD education.