3 Methodology..............................12
3.1 Instruments and Materials..........................12
3.2 Procedures.........................13
3.3 Summary...................................15
4 Results and Analyses............................... 16
4.1 Correlation Analyses..........................16
4.1.1 Indices Significantly Correlated with Public Speech Quality...............16
4.1.2 Indices not Significantly Correlated with Public Speech Quality.........17
5 Discussion..........................29
5.1 Contrastive Analyses of Three Specific Indices of Linguistic Features.......... 29
5.1.1 Pronouns.....................29
5.1.2 Content Words.................................... 30
5 Discussion5.1 Contrastive Analyses of Three Specific Indices of LinguisticFeatures
5.1.1 Pronouns
In the present study, pronoun incidence shows a negative correlation with public speechquality, which indicates that the more pronouns are used, the lower score the speech will begiven. This result is similar to the result obtained by Du[45]. But, the result of this present studyis in contrast to Wang’s[48]. According to Wang, the pronoun ratio is directly proportional tothe machine score. That is, the more pronouns are used, the higher the score. Bai and Wang[57]point out that there is a big difference between manual and machine scoring itself, which ismainly due to the different scoring bases. Considering the surface (case, punctuation, articlelength, etc.) and deep (content perspective, rhetoric, vocabulary, etc.) text features, theprocess of constructing manual scoring text icons is more complicated. In contrast, machinescoring is more mechanical, mainly calculating the frequency of words, the number of targetwords, etc., and comparing the similarity of each aspect with the articles in the corpus toassign scores. In Wang’s study, they used machine scoring. Generally speaking, high-levelwriters tend to use pronouns more densely than low-level writers. High-level writers will usepronouns to complete referential cohesion and avoid repeating the same concept in terms ofvocabulary. As a result, the more pronouns used for machine scoring, the higher the quality ofthe writing will be. It was worth noticing that, in the present study, we used manual scoring.Meanwhile, EFL learners have some confusion about the use of pronouns.
The reason for the different results may be that for public speech, the audience’scomprehension needs to use short-term memory, and the speaker controls the speed of thespeech. The listener cannot review the information as they do when reading to confirm thesemantics of the pronouns. As a result, the excessive use of pronouns can easily causeambiguity, which makes it difficult for the audience to catch the main idea of the speakers’expression. So, the more pronouns are used, the lower the speech score will be.
6 Conclusion
6.1 Major Findings
Public speeches should use simple syntactic structure, not very rich vocabulary, and notmany advanced vocabularies or many cohesive words. Twelve indices of linguistic featuresare found to be correlated with public speech quality. These indices belong to 6 groups:descriptive, text easability component scores, connectives, syntactic complexity, wordinformation, and readability. All of these indices reflect that the word frequency of advancedvocabulary, temporal connectives, narrativity, the content structure, the use of pronouns, andthe modifiers. By calculating the correlation between the linguistic features indices and thestudents’ global scores, the linguistic features of public sp
