The Use of I Think as Different Types of Markers in English Conversations

Abstract

This research investigated the use of I think as different types of markers by Thai learners of English as a foreign language (EFL) to determine whether the use of I think by Thai intermediate- and advanced-level EFL participants showed any significant differences. The types of markers and the functions of I think used by the Thai EFL participants were identified, as were their inappropriate uses of I think. In total, 72 Thai EFL participants were involved in this research, and dyadic English conversations were conducted for data collection. The statistical results illustrated that the Thai intermediate-level EFL participants used I think significantly more often than the Thai advanced-level EFL participants, particularly as a stance marker (SM). By contrast, the Thai advanced-level EFL participants used I think significantly more often as a pragmatic marker (PM) and as a politeness marker (PoM) than the Thai intermediate-level EFL participants. The Thai intermediate-level EFL participants mainly used the SM I think to express a subjective evaluative stance, while the Thai advanced-level EFL participants used I think across a broader range of markers and functions. Their inappropriate uses of I think are also discussed.

1. Introduction

Studies of the expression I think as used by both speakers of English as a first language (L1; Aijmer, 1997; Kaltenböck, 2010) and as a foreign language (EFL; Baumgarten & House, 2010; Huang, 2014) in oral communication have been conducted for decades, as the expression is used frequently by both L1 English speakers and EFL learners, thus playing different roles in oral communication (Diaz et al., 2020; Wang, 2020). Recent studies of Thai EFL learners’ use of the expression I think as a pragmatic marker (PM) have revealed that Thai EFL learners with different English proficiency levels used the PM I think more frequently than they did the other PMs (Pan, 2024) and that the PM “I think was overused by Thai EFL learners” (Pan & Aroonmanakun, 2022, p. 198). However, the results did not fully reveal Thai EFL learners’ concrete uses of the expression I think. Furthermore, whether any uses of the expression I think by Thai EFL learners were inappropriate remains underexplored. As English has been regarded as a lingua franca without necessarily adopting L1 English speakers’ use of English as the norm (Montakantiwong, 2024; Pan, 2024), the presentation of its “frequency information has been over-generalized without considering possible factors in using I think” (Huang, 2014, p. 85).

Accordingly, the aim of this research was to further investigate the use of the expression I think by Thai EFL learners with different levels of English proficiency based on a refined contrastive interlanguage analysis (CIA2; Granger, 2015) without adopting L1 English speakers’ use of the expression I think as the norm. First, this research investigated whether the use of the expression I think by Thai intermediate- and advanced-level EFL learners shows any significant differences in oral communication. Second, it aimed to identify the different types of markers in the use of the expression I think by Thai EFL learners in oral communication. Finally, it aimed to identify whether any uses of the expression I think by Thai EFL learners are inappropriate. Through a fine-grained analysis of the functions of the expression I think, frequency information about the expression I think as used by Thai EFL learners in previous research can be fully understood for both theoretical and pedagogical consideration in the future. The three research questions (RQs) are listed below:

RQ1. Does the use of the expression I think show any significant differences between Thai intermediate- and advanced-level EFL learners?

RQ2. What types of markers are expressed via the PM I think as used by Thai EFL learners?

RQ3. Are any uses of the expression I think by Thai EFL learners inappropriate?

In the remainder of the paper, the previous studies of I think, including the process of its grammaticalization and its various functions as used by both English L1 speakers and EFL learners will be reviewed, followed by the methodology of this research. The results and discussion will be elaborated after the methodology section; the conclusion will be presented at the end of this paper.

2. Previous Studies of I Think

The pronoun I and the verb think can be combined into a subject–verb form (Brinton, 2017). Traditionally, the use of I think that precedes an object clause to provide the speaker’s definite subjective opinion (Farahani & Ghane, 2022). In this case, the use of I think that is a main clause, with a higher hierarchical level compared with the following clause (Aijmer, 1997). However, the subject–verb form I think is currently still being grammaticalized, which refers to the process whereby I think “come[s] in certain linguistic contexts” to “continue to develop new grammatical functions” (Kim, 2024, p. 3). The deletion of the conjunction that creates a linear relationship between the expression I think and its upcoming clause (Brinton, 2017; Huang, 2014). Hence, the expression I think is regarded as a “pragmatic parenthetical” that is “syntactically independent of [its] host or anchor clause” (Brinton, 2017, p. 8). Apart from its syntactic features, the grammaticalization of I think has led to its different pragmatic functions, serving as a range of markers in oral communication.

The first type of marker that I think expresses is a stance marker (SM) due to its “inherent core meaning” that indicates the speaker’s personal opinions (Diaz et al., 2020, p. 512). SMs refer to any lexical items that convey the speaker’s subjective state of mind or attitudinal orientation toward an object (Traugott, 2020). The concept of SMs directly conveys the core meaning of the expression I think, reflecting a subjectivity that involves the speaker’s own opinions. Hence, understanding the frequent use of the SM I think is not difficult because speakers “are not just referring to the outside world but rather are expressing their opinions and attitudes” (Wei et al., 2021, p. 766). The SM marks the epistemic stance, namely “a speaker’s level of commitment to the truth-value of a proposition” (Caprario, 2023, p. 213); the evaluative stance, namely an evaluative judgment of an object by the speaker (Traugott, 2020); and the affective stance, which is an expression of the speaker’s emotion (Mohr, 2021).

The second type of marker that I think expresses is the PM. PMs are lexical items that serve a range of pragmatic functions but have no or little propositional meaning, such as well, I mean (Brinton, 2017; Pan, 2024). The pragmatic functions of PMs can be divided into the discoursal and interpersonal domains. PMs in the discoursal domain mainly connect prior and upcoming utterances, such as the start of a new topic (Crible & Blackwell, 2020; Pan, 2023, 2024), while PMs in the interpersonal domain maintain interpersonal relationships in interactions and indicate thought processes (Schleef, 2023; Tübben & Daniela, 2022). The PM I think is used frequently by both L1 English speakers and EFL learners (Kaltenböck, 2010; Mohr, 2021). In the discoursal domain, the PM I think has been used to mark the start of a new topic, a topic shift, or a summary (Li & Pang, 2022). In the interpersonal domain, it has been used to indicate the thought processes in oral communication and as an approximator. For example, “I think [ninety per cent] that’s what happens” (Kaltenböck, 2010, p. 245).

The third type of marker that I think expresses is the politeness marker (PoM). PoMs are used to indicate an individual’s politeness in interactions, such as please (Kaltenböck, 2010). Since disagreements or conflicts occur frequently in interactions (Miecznikowski & Jacquin, 2023), some lexical items, such as well, I think, can function as a hedge or as a mitigator to ease potential tension or to deescalate conflicts to promote harmonious interpersonal relationships (Pan, 2024). This is, in fact, related to the “tentativeness” in politeness theory as “the avoidance of imposition” in negative politeness, including “indirectness structures and hedge[s]” (Dorst et al., 2024, p. 41).

Table 1. Types of markers of I think.

Markers	Functions	Concise Explanations
stance marker	epistemic stance	indicating certainty or uncertainty
	evaluative stance	indicating evaluative judgments
	affective stance	indicating emotions
pragmatic marker	marking a topic shift	indicating a topic shift
	marking a new topic	indicating a new topic
	marking a conclusion	indicating an end of a topic
	as a planner	indicating the thinking process
	as an approximator	indicating an approximation
politeness marker	as a mitigator	deescalating the conflict

lists the types of markers expressed by I think.

Some studies have investigated I think as a PM by including all the functions listed above (Aijmer, 1997; Pan, 2024). However, based on a nuanced qualitative analysis in several studies (Pan, 2024; Wang, 2020), the diversity of the functions of I think should be separated for a fine-grained analysis. This research does not intend to argue the absolute differences between the terms but rather to divide the types of markers expressed by I think based on different functions determined in previous research to clearly demonstrate its different uses in oral communication. Moreover, previous research indicated that the use of I think in certain situational contexts may be concerned with multiple functions (Diaz et al., 2020; Kaltenböck, 2010). For example, Pan and Aroonmanakun (2022) found that Thai EFL learners used I think for uncertainty since they tended to be indirect in their responses to interactants. In this sense, the use of I think was mainly used as an indication of stance but was also used for politeness. However, a main function of the use of I think exists in each situational context alongside other functions (Pan, 2024).

L1 English speakers use the expression I think not only to convey different stances in different spoken genres but also to introduce new topics or as a mitigator (Huang, 2014; Kaltenböck, 2010). By contrast, EFL learners with different L1s use I think as an SM to elicit different stances significantly more often than L1 English speakers (Kaltenböck, 2010; Pan & Aroonmanakun, 2022). In addition, EFL learners use I think as a PM to indicate thought processes in spoken interactions more frequently than L1 English speakers, while they use it less often for its other functions as a PM compared with L1 English speakers (Aijmer, 1997; Mohr, 2021; Wang, 2020). While several researchers have only examined the use of I think by avoiding a comparison with L1 English speakers’ use, EFL learners with different L1s, including Hong Kong, Thai, Chinese, and Indonesian learners, had their own patterns of using the expression I think, such as an SM to express subjective evaluative judgments (Diaz et al., 2020; Pan, 2024; Pan & Aroonmanakun, 2022; Wang, 2020).

Apart from the results mentioned above, the analysis of the functions of I think and its types of markers in oral communication is still insufficient to understand how Thai EFL learners with different levels of English proficiency use I think in interactions. Its frequency information, particularly the claim of its overuse, may not fully reveal the actual use of I think in situations in which English is regarded as a lingua franca (Montakantiwong, 2024). Even if Thai EFL learners overuse I think, whether the pattern of overuse of I think is the same for Thai EFL learners with different levels of English proficiency remains unknown. Moreover, whether any uses of I think are inappropriate in situations of overuse requires further analysis. An inappropriate use of I think refers to situations in which the function of I think may not be consistent with the given context, such as the redundancy of using I think to mark a subjective stance (Wang, 2020). Several researchers have raised the issue of whether short expressions used as SMs or PMs in a given situational context may have multilayered functions, that is, whether they serve a primary or certain other functions (Kärkkäinen, 2012; Li & Pang, 2022). For example, Kaltenböck (2010, p. 257) found that some instances of the use of I think by L1 English speakers “crucially depend[ed] on its interaction with the linguistic co-text,” since the use of I think could be interpreted as expressing uncertainty or as an approximator; this issue was addressed in the present research. The research methodology is presented in the next section.

3. Methodology

3.1. Participants and Data Collection

This research involved 72 EFL participants whose nationality was Thai and whose L1 was Thai: 36 intermediate- and 36 advanced-level EFL participants at the undergraduate level who were enrolled in years one to four at the same public university in Bangkok, Thailand. They studied different majors, including both the natural and social sciences; they had studied English in Thailand for approximately 13 years and had not been long-term residents in an English-speaking country prior to data collection. Each participant was requested to submit valid scores for an international English exam, such as the IELTS, to confirm that they had the necessary levels of English proficiency based on the Common European Framework of Reference for Languages (CEFR; Council of Europe, 2020). The reason for choosing advanced and intermediate English levels was that the aim of this research was to further investigate the functions of the use of I think following the frequency information provided by previous research on the use of I think by Thai EFL learners with these proficiency levels (Pan, 2024; Pan & Aroonmanakun, 2022). A total of 48 males (67%) and 24 females (33%) participated; however, gender was not a variable in this research.

All participants consented to participation in this study. Each participant was requested to have a dyadic English conversation with another participant whose English proficiency level was the same. Each pair could discuss any topic that they wanted, including friendship, social media, high technology, the natural environment, and social problems. Each pair decided on a specific time after class to have these dyadic English conversations on campus at a location in which they felt comfortable, to ensure that the conversations would be as close to natural as possible despite the data being elicited specifically for this research. Prior to the conversations, each pair was informed that they were not allowed to prepare for the conversations and that English must be used as the primary language; a limited amount of code-switching was allowed if it was essential during the conversation, mainly due to the participants’ deficiency in spoken English. The researcher was not present while each conversation took place to decrease each pair’s nervousness; instead, the researcher audio-recorded the conversations using the Voice Memos Application on an iPhone 13 that was set up before each conversation to collect the data. Each conversation lasted for approximately 20 min.

Finally, the researcher collected a total of 36 dyadic English conversations with an average length of 27.6 min. The researcher used the ELAN Program (2024) for transcription. The ELAN Program (2024) was chosen for the transcription because it allows for multilayered annotations of audio recordings using the Extensible Markup Language (XML) format in Appendix A. The learner corpus entitled “The Corpus of I Think by Thai Learners” (CITTL) that was built based on the written transcriptions contained a total of 92,648 tokens consisting of 35,867 tokens obtained from the 18 dyadic English conversation samples produced by the Thai intermediate-level EFL participants and 56,781 tokens obtained from the 18 dyadic English conversation samples produced by the Thai advanced-level EFL participants.

3.2. Data Analysis

Both quantitative and qualitative methods were used for the analysis in this research. Following the CIA² (Granger, 2015), this research compared the use of I think by Thai EFL learners with different levels of English proficiency without adopting L1 English speakers’ use as the norm. The first step in the data analysis was to identify all the occurrences of I think in a linear relationship with prior and upcoming utterances based on the criteria in previous research (Brinton, 2017). The second step was to identify the primary function of each instance of the use of I think based on Table 1. The identification of the functions of I think was mainly based on the bottom-up approach and a conversation analysis (CA; Clayman & Gill, 2023; Clift et al., 2022). The bottom-up approach is generally used in the field of corpus linguistics to identify the forms and functions in context (Kaltenböck, 2010; Kim, 2024). CA was specifically used to focus on turn-taking in the conversations (Clift et al., 2022). Turn-taking typically occurs in conversation, with each interactant speaking alternatingly. Two raters were involved to ensure high reliability of the identification of all the occurrences of I think and their functions. The inter-rater reliability for the identification of all the occurrences of I think and their functions were 0.996 and 0.968, indicating high reliability. Both raters reached agreement after discussing their different views in certain instances.

In addition, the researcher asked all the participants to identify the functions of each instance of I think that they used in the conversations. The researcher provided the context of each instance of I think with explanations of all the functions of I think based on Table 1 to enable each participant to identify the functions via an online survey. The participants were allowed to identify multiple functions according to their considerations, with a request for a concise explanation. Both raters compared the participants’ identifications with their own identifications to determine whether any differences could be found. The results revealed that the participants layered multiple uses of I think in different situations. Both raters also used the participants’ identifications of the use of I think to identify whether any uses were inappropriate, such as unnecessary uses.

In addition to the qualitative analysis mentioned above, this research used the log-likelihood (LL) significance test to identify significant differences in the use of all the expressions of I think and those with each function in each type of marker to answer RQ1. The LL test is generally used for a comparison of the frequencies of a linguistic item in two corpora (Rayson et al., 2004). The next section elaborates on the results of this research.

4. Results

4.1. Statistical Results

Table 2. Comparison of the use of I think by the Thai EFL participants with both English levels.

Markers	Functions	Intermediate Level		Advanced Level		LL Results
		RFs	NFs	RFs	NFs
SMs	evaluative stance	216	60	95	17	120.19
	epistemic stance	48	13	26	5	20.62
	affective stance	41	11	46	8	2.54
	all	305	85	167	29	129.02
PMs	as a planner	42	12	22	4	18.89
	as an approximator	0	0	20	4	19.58
	marking a new topic	0	0	18	3	17.63
	marking a conclusion	0	0	14	2	13.71
	all	42	12	74	13	0.31
PoMs	as a mitigator	0	0	22	4	21.54
all		347	97	263	46	82.10

presents the raw frequencies (RFs), the normalized frequencies (NFs per 10,000 tokens), and the LL test results for the frequencies of the use of I think by the Thai intermediate- and advanced-level EFL participants, as well as the RFs, the NFs, and the LL test results for the frequencies of the use of I think with each function by the Thai intermediate- and advanced-level EFL participants. According to Rayson et al. (2004), the critical value of the LL test should be 15.13 when p < 0.0001. If the LL test result is more than 15.13 when p < 0.0001, a significant difference exists.

In total, 347 and 263 instances of I think were used by the Thai intermediate- and advanced-level EFL participants, respectively, accounting for 610 instances of I think in the CITTL. The LL test result revealed a significant difference in the use of I think between the Thai intermediate- and advanced-level EFL participants (LL = 82.10, p < 0.0001). In addition, significant differences were found in the use of the SM I think (LL = 129.02.10, p < 0.0001) and the PoM I think (LL = 21.54, p < 0.0001) between the Thai intermediate- and advanced-level EFL participants. As an SM, significant differences were found in the use of I think to express an evaluative stance (LL = 120.19, p < 0.0001) and an epistemic stance (LL = 20.62, p < 0.0001) between the Thai intermediate- and advanced-level EFL participants. Although no significant difference was found in the use of the PM I think between the Thai intermediate- and advanced-level EFL participants (LL = 0.31, p < 0.0001), significant differences were found in the use of the PM I think as a planner (LL = 18.98, p < 0.0001), as an approximator (LL = 19.58, p < 0.0001), and to mark a new topic (LL = 17.63, p < 0.0001). The next section elaborates on the use of different types of markers of I think in detail.

4.2. I Think as Different Types of Markers

As Table 2 shows, both the Thai intermediate- and advanced-level EFL participants used the SM I think the most frequently in the dyadic English conversations, with RFs of 305 (88%) for the Thai intermediate-level EFL participants and RFs of 167 (63%) for the Thai advanced-level EFL participants. By contrast, the Thai intermediate-level EFL participants used the SM I think much more often than the Thai advanced-level EFL participants, particularly to indicate the evaluative stance, as shown in Example 1 below.

(1)	TH02 (00:15:36)
	<TH02 key = “question”>
1	How’s your mom?
	<TH01 key = “response” key = “SM”>
2	I think she’s [ she’s better than [ than uh before.

The participants discussed their family members; TH01 revealed that his mother had recently had an accident and had been in the hospital for a long time up to the day the conversation occurred. TH01 used the SM I think to express his subjective evaluation of his mother’s present condition in Line 2 in response to TH02’s question in Line 1. The evaluative stance was identified via the SM I think having scope over the entire upcoming utterance and the evaluative context (Diaz et al., 2020; Kaltenböck, 2010). By contrast, the use of the SM I think as an epistemic stance marker was not as frequent as its use as an evaluative stance marker by the Thai intermediate- and advanced-level EFL participants. However, the Thai intermediate-level EFL participants still used the SM I think as an epistemic stance marker much more often than the Thai advanced-level EFL participants, as Examples 2 and 3 demonstrate below.

(2)	TH05 (00:12:50)
	<TH05 key = “question”>
1	You sure?
	<TH06 key = “response” key = “SM”>
	<TH06 key = “high-pitched sound”>
2	I think I am sure, very sure.

(3)	TH13 (00:10:28)
	<TH13 key = “question”>
1	OK, what we do about it?
	<TH13 key = “response” key = “SM”>
2	I [ I uh maybe we can uh search Google first, I think.

Some instances of the epistemic stance may not have been particularly different from the evaluative stance due to the semantic sense of the subject–verb form I think proposed in previous research (Aijmer, 1997; Caprario, 2023). However, the prosody and the position of the SM I think differentiated the epistemic stance from the solely evaluative stance, since the speakers intended to display their certainty or uncertainty by uttering the SM I think (Baumgarten & House, 2010). In Example 2, TH05 attempted to confirm a prior question in which TH05 asked for an answer to a problem regarding high technology. TH06 used a high-pitched tone to emphasize the entire utterance in Line 2, indicating that he was not only expressing a subjective evaluative but also deliberately displaying his certainty about the answer. By contrast, in Example 3, in which neither participant knew how to solve a problem with a technical bug, TH13 used the PM uh (Tübben & Daniela, 2022) and the adverbial epistemic stance marker maybe (Caprario, 2023) to indicate her uncertainty and inserted the SM I think at the end of the utterance in Line 2 to confirm her uncertainty (Huang, 2014).

Regarding the use of the PM I think, the Thai intermediate-level EFL participants used it as a planner significantly more often than the Thai advanced-level EFL participants, as demonstrated by Example 4 below.

(4)	TH25 (00:23:14)
	<TH25 key = “question”>
1	You think um friend can be forever?
	<TH26 key = “response” key = “PM”>
2	Uh <pause/3.6 s> I think <pause/3.1 s> um maybe not everyone can.

The participants in Example 4 discussed friendship. TH26 used the PMs uh/um and I think with unfilled pauses to indicate her thought process regarding the prior question (Schleef, 2023; Tübben & Daniela, 2022). The annotation revealed that the unfilled pauses between the PMs uh and I think and between the PMs I think and um in Line 2 were longer than the pauses between other words that TH26 uttered. Although the use of these PMs may reveal EFL learners’ disfluency (Crible & Blackwell, 2020), the context in Example 4 indicated that TH26 intended to think about the answer to the prior question by using the PMs uh/um and I think as planners, as TH26 also confirmed in the online survey. By contrast, the Thai advanced-level EFL participants used the PM I think as an approximator and for marking a new topic, as both examples below demonstrate, while the Thai intermediate-level EFL participants did not use either of these functions.

(5)	TH43 (00:17:52)
	<TH43 key = “question”>
1	You know how much?
	<TH44 key = “response” key = “PM”>
2	I think 2000 dollars, probably more.

(6)	TH66 (00:06:18)
	<TH66 key = “question”>
1	OK, then I will check it out next week.
	<TH65 key = “response”>
2	Sure OK [ OK.
	<TH65 key = “response” key = “PM”>
3	I think we can talk about some social problems now.

Similarly to the findings of previous research, the Thai advanced-level EFL participants inserted the PM I think before numbers, as Example 5 demonstrates, to indicate approximation (Kaltenböck, 2010; Li & Pang, 2022). In Example 5, TH44 used “probably more” in Line 2 to repeatedly indicate that the price he stated was an approximation. In Example 6, after the participants concluded a discussion, TH65 used the PM I think to mark the start of a new topic in Line 3. The Thai advanced-level EFL participants used this function frequently, possibly due to the research setting, in which the participants were allowed to choose their topics of discussion.

Regarding the PoM I think, the Thai advanced-level EFL participants used it as a mitigator, as Example 7 below demonstrates, whereas the Thai intermediate-level EFL participants did not.

(7)	TH54 (00:15:22)
	<TH54 key = “declaration”>
1	I just feel it’s not polite to only think about self.
	<TH54 key = “declaration”>
2	We need to do something for others.
	<TH53 key = “ declaration”>
3	But everyone is thinking themselves.
	<TH54 key = “declaration”>
	<TH54 key = “high pitch, rapid speed”>
4	So you think it’s OK to do bad things to [ to others and only for ourselves?
	<TH53 key = “response” key = “PM”>
5	OK I think this problem we can mark it first, cause you know there’s no right or wrong here.

The participants in Example 7 discussed whether people should contribute more to society rather than being selfish. TH53’s utterance in Line 3 overtly contradicted TH54’s opinion in Lines 1 and 2. This must have annoyed TH54, who used a high-pitched tone and rapid speech to ask the question in Line 4. TH53 must have noticed the potential conflict at that moment, as they used the responsive word OK followed by the PM I think to attempt to deescalate the conflict, stating a temporary resolution with a plausible reason in Line 5 for mitigation purposes (Pan & Aroonmanakun, 2022; Wang, 2020). In addition, besides for the purpose of mitigation, the use of I think in this context also displayed the epistemic purpose of TH53 since TH53 mentioned that “there’s no right or wrong here” in Line 5, indicating TH53’s epistemological subjectivity. Hence, as mentioned in the earlier section, the use of I think may have two functions: here, the main function is for mitigation, while the second function is to indicate the interactant’s epistemics toward the ongoing issue.

Following this detailed analysis of I think with each function as used by the Thai EFL participants, in the next section, we illustrate inappropriate uses of I think.

4.3. Inappropriate Uses of I Think

Two types of inappropriate uses of I think were identified. First, 89 of 305 (29%) and 35 of 167 (21%) instances of the SM I think used by the Thai intermediate- and advanced-level EFL participants were considered to be redundant, respectively, accounting for 124 instances (26%), as Example 8 demonstrates below.

(8)	TH05 (00:11:07)
	<TH05 key = “question”>
1	Oh, where’s the restaurant?
	<TH06 key = “response” key = “inappropriate SM”>
2	I think it’s just in front the university.
	<TH06 key = “declaration”>
3	Just straight in front, very big.

The use of the SM I think is assumed to indicate the subjectivity of a proposition (Aijmer, 1997; Wang, 2020), whereas TH06’s proposition in Example 8 should be an objective fact. Based on linguistic cues, such as “just in front the university” in Line 2, “just straight in front,” and “very big” in Line 3, TH06 attempted to describe the geophysical location of the restaurant to TH05 without indicating any subjective stance toward it. Hence, the use of the SM I think was redundant in this context.

The second type of inappropriate use of I think was related to grammatical errors, as Example 9 demonstrates below.

(9)	TH35 (00:07:36)
	<TH35 key = “declaration”>
1	I play game last night uh until one a.m.
	<TH36 key = “declaration”>
2	OK, because you don’t tell me.
	<TH36 key = “declaration” key = “inappropriate SM”>
3	I think it is too late to call you at nighttime.

The participants in Example 9 discussed their activities from the day before the conversation occurred and used the present tense to describe these events from the past, such as “play” in Line 1. Grammatically, the SM I think in Line 3 should have been the SM I thought. The aim of this research was not to focus on grammatical errors; however, expressions such as I thought could be considered an “epistemic phrase” to “take stances frequently and effortlessly” (Kärkkäinen, 2012, p. 2209). A total of 58 (12%) instances from situations involving the use of I think could be corrected to the SM I thought for taking stances, including 43 (14%) and 15 (9%) utterances by the Thai intermediate- and advanced-level EFL participants, respectively.

5. Discussion

This research investigated the use of I think based on its diverse roles as markers in dyadic English conversations produced by both Thai intermediate- and advanced-level EFL participants from both quantitative and qualitative perspectives. The statistical results indicated that the Thai advanced-level EFL learners chose to use I think significantly less often compared with the Thai intermediate-level EFL learners. This quantitative result can further explain the overuse of I think by EFL learners with different L1s found in several previous studies since the EFL learners involved in previous research had intermediate levels of English proficiency (Huang, 2014; Pan, 2024; Pan & Aroonmanakun, 2022). A closer examination of the roles of I think indicated that the Thai intermediate-level EFL learners used a more limited range of I think, with two types of markers that included four functions, while the Thai advanced-level EFL learners used I think in a broader range of roles and functions, with three types of markers and eight functions of I think. This result suggests that the level of English proficiency was a variable that influenced Thai EFL learners’ use of I think as different types of markers and their functions in oral communication. In addition, the Thai EFL learners with both levels of English proficiency predominantly used the SM I think, reflecting the crucial role of the SM I think in English conversations (Huang, 2014; Wang, 2020). By contrast, the Thai advanced-level EFL learners used significantly fewer instances of the SM I think but significantly more instances of the PM and PoM I thinks compared with the Thai intermediate-level EFL participants, suggesting that the Thai EFL learners with higher levels of English proficiency replied by using I think in the discoursal and interpersonal domains more often than the participants with lower levels English of proficiency. This tendency of the Thai advanced-level EFL learners to use the PM and PoM I thinks is in line with the use of I think by L1 English speakers, suggesting that I think not only elicits subjectivity due to its semantic core meaning but also has a broad range of pragmatic functions in oral communication (Farahani & Ghane, 2022; Traugott, 2020).

With regard to the types of markers of I think and their functions, multilayered functions were identified based on the participants’ annotations in the online survey.

Table 3. Multilayered functions of I think used by the Thai EFL participants.

Markers	Primary Function	Secondary Function	RFs of Intermediate Level	Rfs of Advanced Level	Total
SM + PM	evaluative stance	as a planner	61	45	106
SM + PM	epistemic stance	as a planner	12	8	20
PM + SM	as a planner	evaluative stance	11	9	20
PM + SM	as a planner	epistemic stance	6	8	14
PoM + PM	as a mitigator	marking a new topic	0	9	9
All			90	79	169

lists the multilayered functions of I think with their RFs as used by the Thai EFL participants.

According to the participants’ annotations, the functions of many of the instances that the raters identified as primary functions the participants revealed as secondary functions. Table 3 shows that most of the instances of I think that had two functions had the stance function as an SM and the function of a planner as a PM, as the two examples below demonstrate.

(10)	TH04 (00:25:40)
	<TH04 key = “question”>
1	Do you like him?
	<TH03 key = “response” key = “SM + PM”>
2	I think <pause/1.1 s> he’s a good singer, but he’s not my favorite.

(11)	TH45 (00:18:32)
	<TH45 key = “question”>
1	How’s this?
	<TH46 key = “response” key = “PM + SM”>
2	Um <pause/4.2 s> I think <pause/1.8 s> it looks OK <pause/1.2 s>, very luxury.

In Example 10, the participants discussed a popular singer while looking at a computer screen. The primary function of the use of I think in Line 2 by TH03 was to express a subjective evaluative judgment about the singer. In addition, based on the clause that followed the SM I think and the response to the online survey by TH03, TH03 showed slight difficulty in answering the prior question with only “yes” or “no” at that moment. Accordingly, in addition to using I think to provide an evaluation, TH03 used it to buy some time to think about how to answer the question in Line 1. By contrast, in Example 11, the participants discussed a new product while looking at a computer screen. Based on the co-text, in which the filled pauses uh/um and unfilled pauses were used, this PM I think had the primary function of indicating TH46’s thought process and buying more time to respond to the prior question. Moreover, based on TH46’s response to the online survey, the specific use of I think rather than any other linguistic cues at that moment was also to indicate TH46’s subjective evaluation of the product that both the participants were looking at on the computer screen. Based on these two examples shown above, SMs or PMs, such as well, you know, and I mean (Brinton, 2017; Kim, 2024; Mohr, 2021), were further determined to have more than one function in a given context in oral communication (Kaltenböck, 2010). This phenomenon is mainly due to the polyfunctionality of SMs or PMs in oral communication (Caprario, 2023; Crible & Blackwell, 2020), as well as the indeterminacy of spoken interactions (Pan, 2023, 2024) since speakers do not have much time to prepare upcoming utterances in spoken interactions (Schleef, 2023; Wei et al., 2021).

Lastly, the results also revealed certain inappropriate uses of the SM I think by the Thai EFL participants with both levels of English proficiency. These inappropriate uses of I think may reveal the overuse of I think by Thai EFL learners because the Thai EFL learners tended to insert the SM I think before an utterance that only contained objective facts and focused overly on the use of I think rather than using other, similar SMs that are frequently used by L1 English speakers, including the past-tense form I thought, or I believe and I guess (Brinton, 2017; Kärkkäinen, 2012; Wei et al., 2021).

6. Conclusions

The use of I think in oral communication was explored in depth based on the roles and the polyfunctionality of the three types of markers as used by Thai EFL participants with two different levels of English proficiency. The detailed results of the qualitative analysis complemented the frequency information in previous research. Theoretically, the grammaticalization of I think was reflected by the Thai EFL learners’ use of I think with different functions. However, the types of markers and the polyfunctionality, particularly regarding the PM and PoM I thinks, as well as their discoursal and interpersonal functions, should be included in the pedagogical framework for teaching and learning English when EFL learners’ English levels are not high. Furthermore, the inappropriate use of the SM I think by EFL learners should be paid special attention.

Moreover, the multilayered functions of the use of I think and the other SMs and PMs should be further studied with regard to both L1 English speakers and EFL learners since little research has fully discussed this linguistic phenomenon. The use of I think found in this research may need to be further ascertained by future research involving EFL learners with different L1s and diverse levels of English proficiency.