4.1. The participants’ common errors in the use of FSs
4.1.2. Common FS errors by linguistic categories
In this section, the participants’ FS errors are discussed in terms of their overall counts and relative distribution within each type of FSs. Errors are also explained in consideration of learning factors and the possible sources of such errors. Nevertheless, it is rather difficult to trace any error back to a particular source due to the various mental processes at play. EA researchers have been trying to replicate a representation of the mental processes that learners undergo when making errors (Dulay et al., 1982). Unfortunately, there is no viable method to assess these mental processes as they happen in the mind of the learners. In EA studies, more often than not, the distinctions are made exclusively at the discretion
62
of the researchers in charge due to the fact that mental processes “must be inferred because they cannot be directly observed or measured” (Etzel, 2019, p. 52).
Therefore, what is presented in this study is based on the findings and the interpretations of the researcher to the best of her knowledge. The errors mentioned here might be interpreted differently and attributed to different sources by other researchers who look at them from different perspectives.
It is also important to note that, within the scope of this study, the word error(s) exclusively represents unformulaic word sequence(s) and does not refer to any violations in terms of grammar, spelling, or other linguistic aspects. In this study, FS errors are word sequences that, regardless of their grammatical accuracy, do not conform with the criteria of FSs stated in the operational definition (discussed in Chapter 2, Section 2.1.1 and Chapter 3, Section 3.1). These sequences might be considered correct and formulaic in other contexts with different criteria – but that is not relevant to the current discussion.
As mentioned in Section 4.2.1., from the 20,813-word translation products provided by the participants, a total of 535 FS errors were identified. They are classified into error categories as presented in Table 4.1. The words that cause the sequences to violate the FS criteria, henceforth referred to as inaccurate choice(s), are underlined and marked with an asterisk.
63
Table 4.1. The participants’ FS errors by linguistic categories
Types of errors
Counts
%
Examples
FS errors Standard sequences External Errors
(Inappropriate use of sequences in context)
46 8.58% *full circle common *senses
vicious circle common experiences
Internal Errors (Errors in the components of the sequences) Content words
Verb 200 37.31
%
*feeling difficulties
*set a plan
experiencing difficulties make a plan
Noun 64 11.94
%
emotional
*requirements experience many
*obstacles
emotional needs experience
hardship
Adjective 53 9.89%
*broken sleep
*all-compassing health
poor sleep general health
Adverb 16 2.99%
drawing *out
*continually occupy
drawing up continuously
occupy
Function words
Preposition 101 18.84
%
*in a piece of paper obligations *with
on a piece of paper obligations to Article 21 3.92%
on *the paper take *an initiative
on paper take initiative Others
(Errors in more than one component)
35 6.53%
*intellectual healthy
*each step *each time
mental health one step at a time
The following sections discuss the FS errors made by the participants in this study in the order of frequency. For each type of FS errors, several examples are given and analyzed. In addition, the author of the present study also made an effort to provide probable explanations with regard to the sources of errors. However, as mentioned earlier, it is difficult to determine the exact mental processes responsible for errors. The explanations for error sources provided below should not be taken as definitive or conclusive as they reflect only the perspective of the author of the present study. Alternative interpretations are equally possible and valid.
64
4.1.2.1. Inaccurate verb choices
The most common type of FS errors is inaccurate verb choices. It makes up 37.31% of all FS errors identified in this study. All participants committed this type of FS errors.
For example, some participants translated the phrase giảm nhẹ một hệ quả không mong đợi as *alleviate an unexpected consequence. At first glance, the combination of alleviate and consequence may appear to be semantically compatible. However, it is not considered formulaic by the criteria of this study (FREQ = 10; MI = 2.1). This inaccurate verb choice makes this sequence an FS error. It might have happened because the participant mistakenly assumed that alleviate and consequence make a semantically meaningful combination, and thus, could be used together. In other words, it seems the participant might have formed a misconception that prompted them to produce an FS error. The mental process behind the formation of such misconceptions is termed false concepts hypothesized and could be traced back to the intralingual source (Richards, 1974a).
Inaccurate verb choices have also been found to be the most common type of errors by previous research (Nesselhauf, 2005; Nguyen et al., 2021; Zinkgrọf, 2008). However, in this study, the FS errors involving inaccurate verb choices are somewhat different from the ones reported in other studies. Nesselhauf (2005) found that the nouns action, attitude, problem, or step were the ones causing the most problems for the participants in his study. In contrast, most of the participants in the present study did not seem to have difficulties choosing the right verbs for those nouns. This could be explained partly by the different mother tongues of the participants.
For example, the sequence change your attitude has a very close equivalent in Vietnamese – thay đổi thái độ. The verbs and the nouns in these sequences align with each other nicely, which probably reduced the linguistic burden for the participants. The same could be said about the sequence solve your problem and its Vietnamese translation – giải quyết vấn đề. Although the participants in the present study did not find the nouns action, attitude, problem, or step as problematic as those in Nesselhauf’s (2005), they seemed to struggle with some
65
other nouns such as consequence, habit, and emotion. It appears that learners, regardless of their L1 and the specific nouns causing them trouble, are prone to FS errors involving inaccurate verb choices.
Another aspect where the findings of the present study are different from previous findings concerns delexical verbs. On this aspect, Zinkgrọf (2008) reported that delexical verbs accounted for the majority of the inaccurate verb choices made by learners. However, in the present study, although the participants did make some FS errors in the use of delexical verbs, this type of errors was not common.
Most of the errors involving delexical verbs in the present study might have been the result of false concepts hypothesized, an intralingual process. Firstly, the participants might have assumed that since delexical verbs carry little meaning, they could be used interchangeably. For example, a common error is the unformulaic combination *do+step (FREQ = 2,697; MI = -0.79) instead of take+step (FREQ = 29,589; MI = 3.56). Secondly, it is possible that the participants might have also operated under the false assumption that delexical verbs could work with any nouns. Therefore, they might have used delexical verbs as a failsafe when encountering unfamiliar nouns, for example, *take a therapy (FREQ = 152; MI = -0.58) instead of undergo therapy (FREQ = 346; MI = 5.68), or *take treatment (FREQ = 238; MI = -1.56) instead of receive treatment (FREQ
= 3,002; MI = 3.90). Although the number of FS errors involving delexical verbs is not high compared to other types of errors, they reveal a potential gap in the participants’ knowledge of delexical verbs – an important element in the English language.
The dominance of errors involving inaccurate verb choices might be rooted in the cognitive processes responsible for the acquisition of verbs during language learning. Cognitive linguistic research has demonstrated that the brain is not as good when dealing with verbs as it is with some other linguistic categories such as nouns (Earles & Kersten, 2000; Engelkamp, Zimmer, & Mohr, 1990; Reynolds &
Flagg, 1976). For example, when a new noun is learned, it is easily mapped into a pre-existing category of referents. Therefore, the meanings of nouns are quite
66
stable. In contrast, verbs cannot be mapped directly into pre-existing categories because their meanings are largely context-dependent. According to Kersten and Earles (2004), learners may experience difficulties retrieving verbs from memory for use if the immediate context is different from the context when the verbs are learned. This is because the meanings of verbs are relational to contexts, and thus, have more variations. Such difficulties may also present with other relational categories such as adjectives and prepositions.
4.1.2.2. Inaccurate preposition choices
In line with the prediction made by previous research mentioned in the last section, the second most common type of FS errors identified in the present study is the inaccurate choice of prepositions with 101 instances, making up 18.7% of all errors. This is somewhat similar to the findings of Nesselhauf (2005), Qi and Ding (2011), and Vu (2017). The difference is that the FS errors involving prepositions in this study, despite being the second most common type, are not as pervasive.
In particular, Qi and Ding (2011) found that the wrong choice of prepositions made up 33% of the FS errors in their study, and Nesselhaulf (2005) suggested that prepositions were among the non-lexical items most commonly involved in the errors found in his study. The participants of these studies appear to be comparable – all were English-major students in their third or fourth year of university. The main methodological difference is the forms of data used for error analysis in each study. While the present study employed a translation test to generate data, Qi and Ding (2011) used transcripts of students’ monologues and Nesselhauf (2005) used argumentative essays. In addition, the participants in the present study were allowed to consult dictionaries, which might not have been the case for the other studies. The different L1s of the participants in these studies might have played a role as well. Nevertheless, the data gathered from the test and the questionnaire in this present study was not sufficient to conclude how exactly the difference in L1 could affect the number of FS errors involving prepositions.
A common FS error involving prepositions made by the participants in this study is the use of *in a piece of paper (FREQ = 683; MI = -1.63) instead of on a
67
piece of paper (FREQ = 7,038; MI = 3). This error could likely be attributed to interlingual sources. As discussed earlier, Vietnamese and English preposition systems have different ways of working (Dam, 2001; Vu, 2017). To convey the message write it down on a piece of paper, Vietnamese people could use either viết vào trang giấy, viết lên trang giấy, or viết trên trang giấy. Figure 4.5 shows the popularity of each phrase estimated by Google. Although the preposition trên seems to be the most popular, the other two are also widely used. In this case, the choice of prepositions (vào, lên, or trên) is mostly stylistic and largely depends on the speech habits of individual Vietnamese speakers. There is no established rule that specifies the noun giấy having to go with which preposition. As a result, Vietnamese learners of English may find it difficult to choose the correct preposition to use with a piece of paper, producing several unformulaic sequences such as *in a piece of paper, *to a piece of paper, and *into a piece of paper.
Figure 4.5. Different ways to convey the message “write it down on a piece of paper” in Vietnamese
Another example of FS errors involving inaccurate preposition choices is:
Maladaptive behaviors prevent people *to adapt to either new or difficult
68
situations. In this example, the verb prevent is followed by the wrong preposition.
This unformulaic combination of prevent and to (FREQ = 39,718; MI = 1.81) was used by four participants, suggesting it might be a problem for some learners. This error could probably be traced back to intralingual sources, particularly, the intralingual process of overgeneralization. When learners engage in overgeneralization, they make errors in an attempt to simplify the language system.
Some might apply the rule of one English item to another one that shares some similarities, assuming they work the same way (Richards, 1974a). The participants likely made the error of using prevent and to because they might have believed that the verb prevent should be used in a similar way to the verb allow, which is its antonym. Since prevent is a direct antonym of allow, if the participants had learned the pattern of allow somebody to do something first, they might have applied this rule to use the verb prevent, producing the unformulaic sequence prevent somebody *to do something.
According to Richards (1974b), overgeneralization also occurs when learners combine two L2 structures to create one deviant structure, commonly known as the blending of structures. Another example of this process is the unformulaic use of the noun responsibility and the preposition for as observed in the translations of some participants for the Vietnamese text “Dù bạn có nghĩa vụ gì đối với người khác…”:
…although you might have responsibilities *for others.
Keeping your responsibilities *for someone is good…
Even though you have responsibilities *for someone…
…though you have any responsibilities *for them.
From these instances, it appears that the participants might have known about the two patterns (have) responsibilities to someone and (be) responsible for someone or something. Instead of using either one, for some reason, they might have mixed up these two patterns and ended up with the unformulaic sequence have responsibilities *for someone (FREQ = 18,592; MI = 2.08).
When it comes to inaccurate preposition choices, in addition to overgeneralization, there might be other mental processes at play. A probable one
69
is ignorance of rule restriction. Analogy has been shown to be the major factor behind this process. As Richards (1974a) asserted, “the learner, encountering a particular preposition with one type of verb, attempts by analogy to use the same preposition with similar verbs” (p. 176). In the present study, this process is evident in several instances where the participants used take action *with to translate the phrase đưa ra hành động instead of take action on because in this context, take action might be used similarly to the verb deal which takes the preposition with.
The participants might be familiar with the phrase deal with something; thus, they might have used the same preposition for take action.
Another example is the use of result *to instead of result in when conveying a cause-effect relationship (Patients have a headache when they are sleeping, which results *to a bad sleep). In this case, the participants might have attempted the analogy of the verb lead in the combination lead to. This should be differentiated from the example of prevent *to which was used earlier to demonstrate the process of overgeneralization from allow to. With ignorance of rule restriction, the main factor – analogy – is extended to similar verbs (Richards, 1974a). Take action *with and deal with, as well as result *to and lead to, both convey similar meanings. By contrast, in the process of overgeneralization, the words may be direct antonyms (prevent and allow, as explained above) or have other semantic relationships that might prompt the learners to overgeneralize the rule.
4.1.2.3. Inaccurate noun choices
The third most common type of FS errors is the inaccurate choice of nouns with 64 instances, making up 11.9% of all FS errors. The participants seem to have the tendency to use nouns with similar meanings interchangeably. For example, the phrase nhu cầu về mặt tình cảm was translated as emotional *requirements (FREQ = 3; MI = -1.18) instead of emotional needs (FREQ = 667; MI = 3.08).
This specific FS error might be traced back to communication-strategy-based sources, particularly, the process called approximation. Approximation occurs as learners attempt to use words sharing similar semantic features (synonyms, antonyms, or superordinate items) to replace the words they cannot produce
70
(James, 1998). In this case, the FS error emotional *requirements was probably the result of the participant’s failed attempt at producing the correct noun need.
They might have tried to compensate through approximation, producing the noun requirement. While requirement is a close option to need, the meaning of requirement is more specific and context-dependent. It might be safe to suggest that the FS error emotional *requirements could be the result of approximation via a subordinate.
This frequency of FS errors involving nouns reported in the present study is mostly in line with the findings of Boonyasaquan (2009) and Nguyen et al.
(2021). However, the inaccurate noun choices in the present study are not limited to verb-noun and noun-noun combinations as reported by Nguyen et al. (2021) or adjective-nouns as reported by Boonyasaquan (2009). In fact, in the present study, the wrong choices of nouns cover a range of verb-noun, noun-noun, and adjective- noun combinations, as well as sequences involving prepositions such as *actions of (FREQ = 5,834; MI = 2) instead of acts of (FREQ = 8,874; MI = 3.30).
In summary, the participants in this study committed a wide range of FS errors. The most common types of FS errors identified are inaccurate verb choices (37.1%), inaccurate preposition choices (18.7%), and inaccurate noun choices (11.9%). This is largely in line with findings from previous studies (Boonyasaquan, 2019; Nesselhauf, 2005; Nguyen et al., 2021; Qi & Ding, 2011;
Vu, 2017; Zinkgrọf, 2008).