1. Introduction

1.1. Background

Oral English is considered as the hardest skill to grasp by many Chinese EFL learners compared with the other three aspects of basic language abilities, namely writing, listening and reading. Because the teaching and training on oral English is hugely ignored in most Chinese schools and few resources and chances are available for EFL learners to practice their speaking, there is usually a huge gap for Chinese EFL learners between their oral English ability and the other three aspects of abilities. They make basic grammatical mistakes in speaking when they are even unconscious of that.

To respond to that, a lot of previous research have been done to figure out common grammatical error types committed in EFL learners’ English speaking and analyze the reasons. Some studies using linguistic category taxonomy conclude that tense error is an outstanding problem, and the one dominant reason is the negative transfer of the mother tongue [1-3]. What’s more, based on another grammatical error classification standard, the surface strategy taxonomy, Ting, Mahadhir and Chang in 2010 and Safrida and Kasim in 2016 believe that omission is responsible for the largest percentage of errors [4,5]. Safrida and Kasim state that the major reason is inter-lingual interference [5]. However, the existing research are mainly concerned with error frequencies of different error types and their causes. Few studies have been done to research the relationship between grammatical errors and English-speaking levels. But it is not clear how to affect, which error types are affected, and to what extent it is affected. Therefore, this research focuses on the error frequencies of different types, their differences among Chinese EFL speakers of varying proficiency, and the relationship between the error frequency and EFL learners’ oral levels. The present study collects 74 pieces of 2-minute recordings on an IELTS Speaking part 2 topic from 74 Chinese four-year university students. Their recordings are scored by a former IELTS Speaking examiner and transcribed for marking of grammatical mistakes. Then frequency analysis, correlational analysis is applied. This study aims to identify frequently made grammatical mistakes in the spoken English of Chinese EFL learners as well as the relationship between grammatical mistakes and oral English proficiency levels.

1.2. Theory

The Surface Strategy Taxonomy, which is originally suggested by Dulay et al., is widely recognizable and has been used by many previous studies in this field [4,6,7]. Nevertheless, the classification is too general, so they are usually further categorized according to errors’ linguistic features. Chuang and Nesi improve the version created by Dulay et al. and list common linguistic features for errors under each category based on their empirical error type research [7]. The Taxonomy has been revised, and now has five parts: misformation, misselection, misordering, omission and overinclusion [7]. According to them, an omission error means a missing item; an overinclusion one points to a redundant item; misformation is defined as a mechanical error involving incorrect form of morpheme; a misselection error involves selecting the incorrect form due to a more challenging conceptual activity; a misordering error involves misplacing a sentence’s constituents. The concrete error types under each category are described in Table 1 according to [7].

Table 1: An Improved Version of Surface Strategy Taxonomy [7].

Based on the actual statistics and classification of errors in the present study, some adjustments are made to the chart for the convenience of error description.

1. Changing “Bare singular noun for plural” to “Misuse of the number of nouns”.

2. Changing “Gerund for noun” to “Misselection between gerund and noun”.

3. Changing “Adjective for adverb” to “Misselection between adjective and adverb”.

4. Changing “Verb for gerund” to “Misselection between gerund and verb”.

5. Changing “Gerund for infinitive” to “Misselection between gerund and infinitive”.

6. Adding “Redundant ‘a/an’” as an independent term in the part of overinclusion.

7. Adding “Redundant ‘more’” as an independent term in the part of overinclusion.

Table 2 is the revised table of grammatical error classification.

Table 2: Updated Surface Strategy Taxonomy.

1.3. Questions

1. What are the percentages of four-year Chinese university students with various spoken English proficiency committing common grammatical errors?

2. Is there an effect of grammatical errors on IELTS speaking scores?

3. How is grammatical error frequency related to oral English levels?

2. Literature Review

2.1. The Previous Study about the Grammatical Errors at Home

Numerous existing studies in China show that grammatical errors in second language learners concentrate on tense errors and that the reasons for this are negative transfer from the mother tongue. Chen, by studying the grammatical errors in the classroom discourse of university English teachers, conclude that these grammatical errors are mainly concentrated in tense, coronation, noun singular and plural, and syntactic order, and conclude that these mistakes result from inappropriate transfers to target language from mother tongue. [1]. Accordingly, another study shows that among the spoken grammatical errors of Mongolian English college students retrieved through Antconc, tense errors, noun singular-plural errors, and errors in the use of the crown and personal pronouns are the most common types of errors, with tense errors accounting for the largest proportion, again due to the negative transfer of Mongolian to English [3]. Similarly, when analyzing the grammatical faults in the English writing of high school learners, morphological errors dominated by verb-like errors account for the majority of grammatical errors, and the main causes of such errors are poor grammatical knowledge, overgeneralization and negative migration from the mother tongue [2].

2.2. The Previous Study about the Grammatical Errors Abroad

When it comes to error analysis on the grammar in English speaking, a majority foreign research focus on the frequency of different types of grammatical errors and causes of these errors. Research conducted by Hojati finds that the most frequent grammatical errors are article, clause and preposition. Another research also analyzes the frequency of distinct kinds of grammatical errors in learners’oral English [5]. According to the research, omission occurs the most frequently (55% of the time), thereafter misformation (29%), while addition (13%). Furthermore, inter-lingual and intra-lingual interference are the two main causes of these grammatical errors, with inter-lingual interference accounting for the majority of cases [5]. Besides, the study done by Ting, Mahadhir and Chang in 2010 combines two classification methods, that is, a linguistic description of grammatical errors and the surface structure taxonomy [4]. The findings are as follows: according to the surface structure description, omission along with misinformation constitutes 72% of all errors in grammar, while misordering as well as addition are less common; if descripting errors in a linguistic way, propositional problems and errors in question forming make up 35% of the errors overall, preceding misformation of words and misuse of articles, which each make up approximately 11 percent of the total [4].

By reviewing existing research, it can be concluded that most research in this field have studied the frequency of grammatical errors of different types in speaking, which is understandable as the basis for further research. But sample sizes are relatively small in this research, and few further research has been concerned with the relationship between grammatical errors and English speaking levels, which is one of the research purpose of the present study.

3. Research Design

3.1. Research Objects

Students from Chinese universities in their fourth year are taking part in this experiment. The sample size is 74, mainly from Tianjin University of Finance and Economics, Shandong Normal University and Nankai University, but also some participants from other universities such as Nanjing University and Sun Yat-sen University, with a wide distribution of majors, 53 of which are from English-related students and the remaining 22 from students of other majors.

3.2. Procedures

To ensure that the participants had similar themes to talk about and enough time to show their oral English ability, the question “a positive change” from IELTS Speaking Part 2 was used as the topic, as the questions in Part 1 and Part 3 are changeable and require shorter answers. 74 volunteers were then asked to give one-and-a-half or two-minute answers to the IELTS Speaking Part 2 questions via audio recording or voice calls. In order to simulate real-life responses and to get as close as possible to the actual speaking test, the volunteers were given only two minutes to prepare and five minutes to record or make a voice call.

The data collected were given to an IELTS speaking test pre-assessor for scoring according to IELTS speaking test standard. Besides, the recordings were transcribed for the first time by a speech-to-text software and were then manually checked to ensure the accuracy. To recognize grammatical errors in the recordings, the data collected were analyzed by Grammarly, which had an accuracy rate of 96.83% according to the Grammarly official website. To further improve the accuracy, the errors were checked and identified manually for a second time by the researchers. After that, the data were categorized and tagged based on the categorization standard mentioned above in order to calculate the frequency of each grammatical error, before being imported into SPSS software for correlation analysis to look into the relationship between grammatical errors and test scores.

4. Result

Based on a summary of the data, researchers found that some grammatical errors had a frequency of 0, including 1.4, 1.9, 2.4, 2.7, 2.12, 2.13, 2.15, 2.17, 2.21, 3.3, 4.1 and 5.2. In addition, researchers counted the grammatical errors with a frequency greater than 20 and calculated they’re of the total number of errors (calculated to two decimal places) (see Table 3).

Table 3: Aggregate table.

From Table 3, people can see that there are six common grammar errors in IELTS speaking: Missing preposition, Missing ‘a’/ ‘an’, Misselection between prepositions, Misselection between verb base form, past participle and present participle, Quantifier/Determiner-noun non-agreement in number and Redundant definite article.

After removing the data with zero occurrences, the summary tables were imported into SPSS software for analysis. Firstly, a normality test analysis was performed, setting the original hypothesis p that the data conformed to a normal distribution, with the following results (see Table 4).

Table 4: Normality test.

* p<0.05 ** p<0.01

All of them demonstrated significance (p<0.05), indicating that the original claim that the normal distribution was not satisfied was not true, so the Spearman correlation coefficient was used in the correlation analysis and the results were as follows (see Table 5).

Table 5: Spearman Correlation.

* p<0.05 ** p<0.01

The researchers were able to draw the results from Table 5: The hypothesis is that there is a significant positive correlation between the scores and 1.6 with a correlation coefficient of p1, p1=0.255, showing a significance level of 0.05. The correlation coefficient between the hypothetical score and 2.6 is p2, p2=0.268, showing a 0.05 level of significance and a significant positive correlation between the two. The remaining items were not significantly correlated.

The quantity of grammatical errors and speaking scores were found to be correlated. Prior to that, the normality test revealed that none of them were normally distributed, and Table 6 shows the results of the Spearman correlation test.

Table 6: Spearman correlation.

* p<0.05 ** p<0.01

All p-values are greater than 0.05, demonstrating that there is no link between the number of errors and the score. This indicates that there is no statistically significant relationship between the number of errors and the score.

5. Discussion

Based on results above, it is found that the six most frequent error types are: misselection between prepositions, redundant definite article, missing indefinite article, quantifier/determiner-noun non-agreement in number, missing proposition and misslection between verb base form, past participle and present participle. Misselection and omission are two prominent errors in the present research, such as missing “to” in “I want talk about”, using “gain” instead of “gained” in “I lost some weight and gain tense muscles”. Similar result is shown in the research done by Ting, Mahadhir, Chang in 2010 [4]. It is also worth mentioning that misselection is regarded as part of misformation in categorization methods of that previous article. From the perspective of linguistic features, errors related to preposition, including omission and misselection, account for the largest proportion of errors. For example, “for” is omitted in “I never asked help”; “to” should replace “for” in “here is the answer for the question”. Ting, Mahadhir and Chang also conclude that preposition is the most frequent error type based on linguistic description, while Hojati concludes that preposition is the second most frequent error type following article [4]. The reasons why EFL learners are vulnerable to make preposition errors in their speaking lie that firstly, English prepositions feature complexity and they may convey different meanings in different contexts, so it is hard for foreign language learners to grasp and apply all the meanings of a particular preposition accurately; secondly, because prepositions of one language scarcely have totally equivalent counterparts in another language, EFL learners’ understanding on prepositions based on their mother tongue may interfere with their learning of English prepositions [8,9]. Another outstanding problem exists in the incorrect use of articles, including missing indefinite article and overuse definite article, such as missing “a” in “I changed in positive way” and using “the” unnecessarily in “fields like the public speaking”. The omission of articles is believed as a kind of inter-lingual error, because the mother tongue of EFL learners, which is Chinese in the present study, does not put an article before a noun in most cases. For the redundant use of definite articles, both intra-lingual and inter-lingual factors take some responsibilities. Students themselves do not have a complete command of the usage rules, and on the other hand, English definite articles cannot find counterparts in the learners’ mother tongue, making the rules challenging and confusing to learn [10].

When it comes to the relationship between grammatical errors and EFL students speaking levels, which are manifested by their IELTS speaking scores, the researchers have studied the relationship between the total error number of subjects and their scores, as well as the correlation between each particular error type and the scores. It is found that there is no clear relationship between how many errors one commits and how high the score he or she gains. Previous research on that aspect can hardly be found, except that Ting, Mahadhir and Chang conduct a similar research between the two elements in 2010 [4]. They analyzed the changes in grammatical accuracy through the performance of the University of Malasia students in spoken English over the course of a semester and eventually found that grammatical accuracy increased [4]. As for the reasons for result differences between the previous research and the present study, firstly, the subjects in the previous research are low-level English learners while the ones in the present research cover nearly every level of English learners. Those less proficient university students may progress from low levels of English speaking to intermediate levels after a whole semester, so they cannot represent the whole trend of grammatical accuracy from a beginner’s level to a professional level. What’s more, the previous study is diachronic while the present one is synchronic, so the subjects remain the same in the former one but differ in the latter one if changes in English speaking levels are focused on. Thirdly, the two groups of people are studied in different contexts. Malaysian university students in the previous study have enough time to prepare their class presentations and consciously avoid some grammatical errors by repetitive practices, whereas Chinese IELTS testers are more likely to improvise to make the recording. On the other hand, to research the correlation between specific error types and EFL learners’ speaking levels, significant positive correlations are shown between test scores and “missing auxiliary ‘be’”, as well as between test scores and “noun for adjective”. For instance, “am” lost in “what I thinking right now”; “confidence” is misused in “the more confidence I am”. As the scores increase, more errors of “missing auxiliary ‘be’” are found, mainly in sentences in a passive voice, present processing tense, or predicative structure. As learners’ English levels increase, they tend to apply more complex voices and tenses in their speaking. As a result, they run bigger risks of committing errors in the usage of auxiliary “be”. EFL learners are also prone to lose “be” before adjectives which can serve as verbs in other circumstances, so they misunderstand the part of speech and assume the auxiliary “be” as unnecessary. Similarly, as the learners’ grades are growing, they expose more errors in their misuse of nouns for adjectives, such as “silence” for “silent”, “regret” for “regretful”. By contrast, the misuse of adjectives for nouns is rarely seen. Because in most cases, students are firstly taught the nominal form of a particular word in the classroom and thus have the deepest impression on it. Consequently, they tend to use the noun instead of the correct adjective form to get their meaning across quickly when they fail to think twice. To explain why the error frequencies, grow as learners’ oral English levels improves Brown points out that applying communication strategies such as avoidance is a significant factor to affect error commitment [11]. Lower-level English speakers are more likely to avoid using complicated or challenging grammatical structures and are rarely exposed to these structures, so they have fewer opportunities to face such challenges and show their disadvantages. However, as learners make progress, they use these grammatical patterns more and thus expose their defects more.

In terms of contributions or inspirations of the research, it plays a role in helping EFL teachers and students view oral grammatical errors that students make from a vantage point and develop a deeper understanding on them. The present study analyzes the causes for the frequently committed errors and the error frequency changes as students speaking levels improve, as well as emphasizing the following difficult language points: usage of prepositions and articles, omission of auxiliary “be” and misuse of noun forms for adjectives. Based on that, language teachers can lay more emphasis on the teaching and practicing of these language aspects and pay more attention to students’ use of auxiliary “be” and adjective forms as they are progressing.

Limitations of the present research are mainly shown in the following aspects. Firstly, the sample size is not desirably large enough and sample numbers in each score level are not very average. Secondly, the chosen speaking topic “a positive change in your life” induces subjects to use the past tense but to nearly lose the opportunity to use the third-person singular form, where many learners may commit the error of subject-verb non-agreement. Therefore, it is difficult and impractical to expose all the potential error types a student can make. More topics are needed to ensure the comprehensiveness of the research. Thirdly, based on the fact that few former research is found in regard to the relationship between error numbers and speaking levels. It is not convincing to conclude that error numbers and English-speaking levels are uncorrelated solely based on the present study. More research in this field is needed to justify the conclusion.

For future research, some improvements can be made in sample sizes and their distribution in different speaking levels. The quality of samples can also be advanced by providing more topics and strictly limiting the subjects’ preparation and talking time. What’s more, future research can make more efforts to figure out the relationship between error totals and speaking levels, as well as some particular error types and speaking levels, because conclusions in these field can hardly be found nowadays.

6. Conclusion

The present study finds that the misuse of functional words are outstanding grammatical errors made by Chinese EFL learners, which is characterized by the omission and misselection of prepositions, omission of indefinite articles and redundant use of definite articles. The reasons why learners are more likely to misuse prepositions and articles are that there are large differences between English and their mother tongue in these two types of words, and that the intricacy and complexion of their rules should also take some blame. What’s more, the researchers find that there is no correlation between error numbers and English-speaking levels, which needs to be further testified due to the lack of related research. In addition, it is concluded that two types of errors present a positive correlation with the subjects’ IELTS speaking scores. There are “the omission of auxiliary ‘be’” and “noun for adjective”. Because students try to use more complicated grammatical patterns as their language levels improve, and they have more chances to commit such errors. These findings inspire language teachers to pay more attention to the above-mentioned parts of speech and make clearer illustrations for some confusing rules. The present research also has defects that the samples are not satisfactorily in large numbers and high quality. Research related to the correlation between speaking levels and error numbers, as well as between speaking levels and error types are needed to testify and further develop the findings made in the present research.