Prosodic parameters of reading in 2 nd to 5 th grade students Parâmetros prosódicos de leitura em escolares do segundo ao quinto ano do ensino fundamental

Accepted: September 17, 2017 Study multicentric conducted at the Universidade de Brasília – UNB Brasília (DF), Brasil based on data collected in Minas Gerais state, Brazil. 1 Universidade de Brasília – UNB Brasília (DF), Brazil. 2 Centro Universitário Metodista Izabela Hendrix Belo Horizonte (MG), Brazil. Financial support: This study was partially funded by CNPq (477155/2013-8) and through undergraduate research scholarships granted by Fundação de Apoio à Pesquisa do Distrito Federal (FAP/DF) (0193.000136/2015). Conflict of interests: nothing to declare. ABSTRACT


INTRODUCTION
Research on reading fluency and its relationship with school development has been growing over the years, and the oral reading fluency is an important tool to monitor this evolution (1) .The way prosody is used in reading directly reflects the fluency of readers, and demonstrates their ability to convey and understand information.
Although the concept of prosody is often limited to intonation or accentuation, it is still a subject of discussion among several scholars.Regarding its formal aspects, prosody is composed of parameters phonetically described as intensity, melodic variation, and temporal organization of speech.Intensity refers to the amount of energy present in the vibratory movement, and it is represented by dB (decibel); melodic variation is a recurring measure that characterizes the melodic curve, and is associated with fundamental frequency (F0), which is described as the mean frequency of vocal fold vibration and is represented by Hz (hertz) (2) ; the temporal organization of speech includes pause, elocution time, and total articulation time, and through these measures it is possible to obtain others, such as the rates of speech and reading (2,3) .
In a study involving children with dyslexia and a control group, prosodic aspects were compared in the reading of such children, including the aspects of melodic variation.From the values obtained by the rates of elocution and articulation, slower reading speed and slowness in the production of each articulatory gesture were observed in children with dyslexia compared with those of children in the control group.This slowness reflects in the characteristics of melodic variation (4) .
A study conducted with 32 5 th grade students, after stimulation based on prosodic patterns, concluded that temporal aspects significantly affect reading fluency (5) .
A longitudinal intra-subject study, which aims to characterize the growth in prosody between 2 nd and 3 rd grade students, demonstrated that control of reading speed is directly associated with prosody, although expressiveness is as necessary as speed (6) .
Therefore, considering melodic variation and reading speed as prosodic parameters determining for reading fluency, the present study aims to characterize and compare the reading of Elementary School (2 nd to 5 th grade) students in relation to the aforementioned aspects and to verify their evolution throughout the schooling years.

METHODS
This study was approved by the Research Ethics Committee of the aforementioned Institution under protocol no.CAAE: 38861914.4.0000.5096.All participants and their parents/guardians signed an Informed Consent Form (ICF) prior to study commencement.
Study participants were 79 Elementary School (2 nd to 5 th grade) students enrolled in the private education network of Belo Horizonte, Minas Gerais state, Brazil.One student was excluded due to acoustic interference in the collected audio, thus 78 schoolchildren participated in the study: 20 from 2 nd grade, 12 from 3 rd grade, 23 from 4 th grade, and 23 from 5 th grade.
The students presented no speech, hearing, and written language disorders.To confirm these data, two questionnaires addressing issues associated with these aspects were applied, one to the children's legal guardians (7) and another to their teachers (8) .
The text "A coisa" (The Thing) (9) , which is a simple text with possibilities of prosodic modulations and easy decoding, was used for data collection.The text was initially read silently and then aloud, according to the children's initiative.
For analysis of melodic variation, the beginning and the end of each audio recording were demarcated using the acoustic analysis program Praat 5.4.19 in order to determine the exact duration of each reading.After that, mathematical treatment was applied to the audio recordings with the objective of eliminating macroprosodic effects by means of the wave smoothing algorithm of fundamental frequency (F0) (10) .
The F0 curve was used as the basis for prosodic analysis.The maximum and minimum points of F0 of the reading of the full text were selected, and the minimum point was subtracted from the maximum point, resulting in the melodic variation.The same procedure was adopted for analysis of the title and expressive phrase (short phrase with an exclamation mark).
For analysis of reading speed, the number of words read (WPM) and the number of correct words read (CWPM) per minute in the full text and the 1 st minute of reading were determined.Statistical analysis was performed using descriptive statistics (mean and standard deviation) and the Student's t-test was applied to the two samples at significance level of 5% for comparison of the results.

RESULTS
Assessment of melodic variation shows an increasing trend of this prosodic parameter throughout the schooling years, especially during 5 th grade (Table 1).The same pattern can be observed with respect to reading speed (Table 1).
For analysis of melodic variation, readings of the title, expressive phrase, and full text were compared between students from 2 nd to 5 th grades, as shown in Table 2.
The aforementioned variables were assessed in order to compare the performance of the classes studied.Statistically significant difference was observed regarding melodic variation only in the comparison with the 5 th grade.In contrast, the reading rates (CWPM and WPM) showed statistically significant differences for practically all comparisons.
In the comparison between the number of words read in the 1 st minute of reading and WPM, values of 0.8, 0.6, 0.4, and 0.6 were found for the 2 nd , 3 rd , 4 th , and 5 th grades, respectively.

DISCUSSION
The variables Full Text, Title, and Expressive Phrase were considered in the assessment of melodic variation.Analysis of these variables showed evolution of Full Text and Expressive Phrase in the comparison between the reading of 5 th grade students with that of students of the other grades.As for Title, no significant changes were observed in the comparison between grades (Table 1).Only the variable Full Text presented statistically significant difference between the grades studied (Table 2).In the comparison of each variable between the grades investigated, difference was observed regarding the reading of the full text only during 5 th grade (Table 2).
In view of this, it is possible to conclude that the full text seems to be the best variable of melodic variation to assess prosody in oral reading, and that differences are relevant only during 5 th grade.Studies have shown that in the comparison of reading aloud between dyslexic children and a control group, considering the prosodic parameters of F0 (maximum, minimum, tessitura, and reading speed), typical readers produce melodic variations and reading speed more appropriately than children with dyslexia (11,12) .
For the analysis of reading speed, the variables WPM, CWPM, and 1 st minute of Reading were considered.For the analysis of WPM and CWPM, a comparison of such parameters was conducted between all the grades studied, and statistically significant difference was observed in practically all the grades confronted, except in the comparison between the 3 rd and 4 th grades (Table 1).
This demonstrates that the performance of oral reading and comprehension improves as schooling progresses (13,14) .
No statistically significant differences were observed in the comparison between the rates of WPM and 1 st minute of reading (Table 2).Therefore, it can be concluded that it is possible to analyze reading rate by counting only the number of words read during the first minute, obtaining a result similar to that of the analysis of WPM.

CONCLUSION
Based on the data of this study, we conclude that, melodic variation (F0) increases as schooling progresses, mainly during 5 th grade.With respect to reading speed (Duration), the number of words read as well as the number of correct words read per minute tend to expand throughout the schooling years.

Table 2 .
Comparison between the variables found for melodic variation (full text, title, and expressive phrase) *Values with statistically significant differences Caption: Expressive phrase: a short phrase with an exclamation mark

Table 1 .
Mean and standard deviation of the results found regarding melodic variation (full text, title, and expressive phrase) and reading speed (CWPM, WPM, and 1 st minute of reading) for each grade assessed (2 nd , 3 rd , 4 th , and 5 th ), as well as the comparison between them Values with statistically significant differences Caption: CWPM: Correct words read per minute; WPM: Words read per minute; 1 st MINUTE: Number of words read during the first minute of recording; SD: standard deviation *