Introduction
Phonological skills are a vital aspect of language development. They are associated with the maturity of literacy skills at school. Evaluating children's phonological skills is crucial from three perspectives: determining the basic level of the child's phonological skill, choosing the goals of the intervention, and examining the effect of the intervention on the child. Quantitative and qualitative methods can be used to evaluate children's phonological skills. Quantitative methods, as an indicator of the severity of the child's phonological problem, allow comparing the children with other groups and are useful for monitoring the intervention results [1, 2]. The phonological mean length of utterance (PMLU) is a quantitative scale to assess phonological skill. PMLU has been proposed as a diagnostic and clinical criterion in phonological development, which can also show the effectiveness of the intervention [3]. PMLU is a word-level scale and is calculated out of the utterance sample of children based on the phonemicized words. This scale consists of two parts which are calculated by scoring consonants and vowels in words produced by the child as well as the target words in the production of the criterion of native-speaking adults. It evaluates four characteristics of children's phonological development. These four features are proximity, complexity, correctness, and variability. 
Proximity is a reflection of the correctness of the word, and when the child produces the word correctly, the score of the proximity ratio of the whole word becomes one, indicating the correctness of the production. The purpose of complexity is to pay attention to the word's syllabic structure; words with monosyllabic and disyllabic structures are considered words with simple structures, and words with polysyllabic structure are complex words. Correctness means the correct production of all components of the word chain, which includes consonants and vowels. Variability refers to how children produce words independent of their phonological form [4]. In addition, the proportion of whole-word proximity (PWP) is another quantitative phonological scale calculated based on the ratio of PMLU of produced words to the target and indirectly indicates speech intelligibility [5]. Because of differences in syllabic and phonemic structures in different languages, this scale should be studied for each language. The PMLU scale was specifically designed to assess spontaneous speech. 
Spontaneous speech evaluation methods are valuable because they show the natural growth of word complexity and focus on the effect of morphological and syntactic skills, along with increasing the length and complexity of words [6]. Because the PMLU has not been investigated in the Persian language and contradictions are observed in studies regarding the sensitivity of this scale to growth, this study aims to obtain PMLU and PWP values ​​in Persian-speaking children aged 48 to 60 months with the Isfahani accent (Isfahani accent is an accent talked by people living in Isfahan City, Iran). We compared PMLU and PWP values in two types of story narration and conversation sampling; also, we examined the sensitivity of the PMLU scale to growth.
Materials and Methods
The current observational study was carried out for one year by an analytical cross-sectional method in Isfahan City, Iran, in 2015. The study population included 100 children (49 girls and 51 boys) in the story narration group and 67 children (32 boys and 35 girls) in the conversation group with an age range of 48 to 60 months selected by the convenience sampling method. With the cooperation of the teacher and referring to the children's educational and health files, the speech and language pathologists in each kindergarten examined the children for the inclusion criteria. The Ages And Stages Questionnaire, second edition, was used to screen children [7]. The inclusion criteria included an age range of 48-60 months, speaking the Persian language, and being monolingual. The exclusion criteria included non-cooperation of the participants for the evaluation or having a history of speech, language, neurological, and hearing problems. Finally, normal children in terms of developmental monitoring criteria were selected. Sampling was done by using the storybook of Shangul-o-Mangul [8]. Most children were familiar with this story, but some were not, so the researcher told them the story. To eliminate the effect of direct learning, the child would ask to retell the story after 5 minutes. After that, to extract the conversation sample, we used topics such as a birthday conversation sample, a memory of a day outing in the park, a description of the characteristics of a good girl and a good boy, and a memory of a trip with parents [8]. Clear Record Lite v. 2.1 was used to record the voice of children in a room. The first 50 words of each speech sample were phonemicized, and finally, the target PMLU and child PMLU values were calculated based on the speech of Persian-speaking adults with an Isfahani accent and the production of each child, respectively. The calculation rules were considered based on the rules provided by Ingram [4]. For example, if the child said the word/baba/(as the correct pronunciation in Farsi, meaning father) correctly, to calculate the child's PMLU, the child would receive one point for each consonant and vowel and one additional point for each correct consonant. (4+2=6). If the child produced the same word as/bapa/, one point would be deducted for each consonant that the child mispronounced (4+1=5). To calculate the target PMLU, the same calculation rule was used, with the difference that for each word, the full points of vowels and consonants were calculated. In addition, for each language sampling method, the mean PWP score was obtained by dividing the mean PMLU score of the child by the mean PMLU score of the target. To examine the agreement between the evaluators, each of the evaluators randomly re-phonemicized 20% of the speech samples of the story and calculated the mean of both PMLU (child and target). The participant's scores were entered into SPSS 16 software separately for stories and conversation. To examine the data distribution, the Kolmogorov-Smirnov test was used, and according to the data distribution, paired t and Wilcoxon tests were used to compare the scales, and the Pearson and Spearman tests were used to examine the correlations.
Results
 Story narration and conversation 
A total of 100 samples were analyzed in the story narration section, and 67 samples were analyzed in the conversation section. Among the 100 samples, 49% were in the age range of 54-48 months, and 51% were in the age range of 55-60 months; however, in the conversation, these age ranges accounted for 52.20% and 47.80%, respectively. According to the sampling method of story narration, the child PMLU, target PMLU, and PWP scales were 8.794, 8.811, and 0.998, and the conversation was 9.068, 9.093, and 0.998, respectively. Table 1 presents the central and dispersion indices of child PMLU, target PMLU, and PWP scales separately from the two sampling methods of story narration and conversation. 


Correlation of phonological mean length of utterance (PMLU) and proportion of whole-word proximity (PWP) with age 
Table 2 presents the results of investigating the correlation of age (in months) and PWP and PMLU scales in two types of story narration (normal distribution) and conversational (non-normal distribution) sampling.


These results showed a significant correlation between age and PWP scale of story (r=0.308) and conversation (r=0.313). 
Comparing the results of story narration and conversation samplings 
The Kolmogorov-Smirnov test was used to check the normal distribution of the data. According to the non-normal distribution of the child’s PMLU scale and PWP, the non-parametric Wilcoxon test was used to compare the child’s PMLU in the two sampling methods of story narration and conversation. The results showed a significant relationship between the child’s PMLU scale in the two sampling methods (P<0.001), but PWP did not show a significant difference between these two sampling methods (P=0.973). According to the normal distribution of the data, the result of the paired t-test to compare the target PMLU in the two sampling methods of story narration and conversation indicated the significance of this comparison in the two types of sampling (P<0.001). The agreement between the evaluators was 70%. 
Discussion 
The first objective of this study was to determine the values of PMLU and PWP in Persian-speaking children aged 48 to 60 months with an Isfahani accent in story narration and conversation. The child PMLU was 8.794 in story narration sampling and 9.068 in conversation sampling, which is lower compared to Canadian-speaking children in the same study and age range [6].
The characteristics of the mother tongue influence the PMLU scale, and different accents may affect it, too [9]. This scale has been studied in different languages ​​so far. The reason for different PMLU studies on different languages, besides the differences in the languages, can be due to the lack of homogeneity of the tasks used and different sample sizes. In other words, the reason for the differences observed in studies in different languages ​​is not completely related to language differences which should be considered in future studies [1, 3, 4, 5, 6, 10, 11, 12, 13]. 
The second objective of this study was to investigate the correlation of age with PMLU and PWP scales in 4 to 5 year-old Persian-speaking normal children with an Isfahani accent. In examining the correlation of age with the child’s PMLU (P conversation=0.276 and P story narration=0.427) and target PMLU (P conversation=0.313 and P story narration=0.493), it can be seen that PMLU has no significant relationship with age. This result is consistent with Kumar’s results regarding slow phonological development in the age range of 4-5 years [6].
However, the PWP of the story narration and the conversation show a significant relationship with age. This relationship is such that with increasing age, PWP increases both in conversation (P=0.032) and in story narration (P=0.030). PWP indicates the stability between the child’s productions and the target and shows the closeness of the child’s productions to the target words with what adult speakers pronounce them. The phonological acquisition does not occur only because of the increased complexity of words but rather because of the child’s need to constantly maintain the connection between his productions and the target [8]. However, this relationship in our study shows a weak power because age can only explain 0.094 of the increase in story narration PWP and 0.097 of the increase in conversation PWP. Therefore, to achieve a more reliable result, it seems necessary to conduct more studies with a larger sample size. 
The last objective of this study was to investigate the effect of the sample extraction method on these scales. By comparing target PMLU in story narration and conversation (P<0.001), we found a significant relationship between the child’s PMLU in story narration and conversation showed (P<0.001) but no significant relationship between PWP in story narration and conversation (P=0.973).
Among the studied studies, except for Ingram’s study [4], no study had investigated two types of sampling tasks for evaluating the PMLU scale. This study examined the PMLU scale in the expression of a single word in the Goldman-Fristoe production test compared to the conversation in a child with a phonological disorder. We cannot compare this study with the present study due to the lack of similarity in the type of speech sample and the participants. 
The reason for the significant difference in the sampling type of story narration and conversation in examining the PMLU scale may be related to the fact that story narration, compared to the conversation, faces the child’s language skills with challenges and difficulty. Wagner et al. also referred to this issue regarding the sampling of story narration. They also pointed out that the clarity and fluency of speech in the conversation are higher than in story narration; the reason is the processing load and more planning of story narration [14]. On the other hand, it can also be pointed out that the sampling type of story narration is more structured and less natural than conversation, and as a result, the child has less freedom to use lexical, phonics, and phonological skills. In the narration of the story, children use similar words due to the imposed framework of the story, but in conversational speech, the children express their skills more freely according to their linguistic abilities. In addition, children use more diverse words in conversation due to being freed from the inflexible frameworks in the story. Sutwood’s study also believes in these differences [15].
Researchers can use the present study as a preliminary study in the quantitative phonological assessment of children. However, the data obtained from the study for the use of therapists, due to the cross-sectional nature of the study, the investigation in a limited age range, and the Isfahani accent, is suggested as purely descriptive. In addition, it is suggested that in future studies, PMLU should be investigated in wider age groups, languages, dialects, and common dialects of Iran and different socioeconomic and cultural situations. It is also recommended to examine this scale in different speech samples and compare their results to ensure the effectiveness of the sampling type on this scale. On the other hand, it can be effective to use this scale in clinical groups and comparison with children without disorders to determine possible differences and to answer whether PMLU can be used for diagnostic purposes. 
The results of this study and other similar studies can be summarized as follows. Because each language has unique characteristics, such as the complexity of words and special phonological rules, it can affect phonological scales, such as PMLU; therefore, it is not far from what is expected to be said that PMLU is influenced by native language characteristics. On the other hand, the different socioeconomic and cultural conditions in each country and language also affect phonological development via the effect of the type of words used.
In examining the relationship between age and PMLU, due to the contradictory results, it seems necessary to examine different age ranges in different studies and methods of articles for a comprehensive and fundamental review of different age ranges. Also, conducting a longitudinal study to observe PMLU changes during growth can be useful in clarifying the relationship between this phonological scale and growth.
 In examining the effect of the type of sampling on PMLU and PWP scale, the lack of a significant relationship between PWP in story narration and conversation indicates that if the PWP scale is used in clinical work, the difference in the type of speech sample (narration of story or conversation) does not cause a difference but when the desired scale is PMLU, the difference in the type of sampling will be effective and crucial.

Ethical Considerations
Compliance with ethical guidelines
This study received the code of ethics (with reference number 295089) from the Research Institute of Primary Prevention of Non-communicable Diseases. The participants’ parents completed the informed consent form for their children’s participation in the research. Participation in the study was completely voluntary, and if they did not want to cooperate, they could be excluded from the study. In all stages of the research, ethical considerations were observed. All the obtained information, including name, specifications, and evaluation results, were reserved. The evaluations did not have any adverse effect on the studied population

Funding
This research was conducted with the financial support of the Research Institute for the Primary Prevention of Non-communicable Diseases. This article was derived from the undergraduate thesis of Fatemeh Karimian, Roya Mohammadi, Zahra Bemani, and Narges Taheri in the Speech Therapy Department of Isfahan University of Medical Sciences. 

Authors' contributions
Project management, conceptualization, method, and validation: Franak Kianfar and Yalda Kazemi; Study analysis: Franak Kianfar, Yalda Kazemi, and Fateme Karimian; Research and review: Fateme Karimian, Roya Mohammadi, Zahra Bemani, and Franak Kianfar; Sources: Zahra Bemani and Fateme Karimian; Draft writing: Roya Mohammadi and Fatemeh Karimian; Finalization of the text, editing: All authors. Supervision and funding: Franak Kianfar.

Conflict of interest
The authors declared no conflict of interest.

Acknowledgments
We are grateful to Narges Taheri, who helped us collect and analyze samples.


References

Appendix 1 
Instructions and rules
1. Calculate the clear words in the audio sample of the child’s speech. 
2. Remove children’s words that are not used in adult speech. 
3. Remove the expressions of the format “There was one, there was no one under the blue dome, there was no one but God.” 
4. Remove the words out of the course of the story that the child produces, such as: Let’s go to the next page, and so on. 
5. Count the words once the child repeats them exactly during the sample. 
6. Do not count the words in which the heart process (Qayem...Qami) ​​occurs. 
7. If the child corrects his speech, calculate the corrected items. 
8. The glottal stop should only be written, but it is not given a point in the calculation as a consonant. 
9. Do not count phonemes like “/Taq-Taq/.” 
10. Do not count the words that carry the child’s emotions, such as/Oh/, /Hurrah/, /Wow/, etc. 
11. Do not count grammatical errors that affect the word form, such as: go, sew, etc.
12. Pay attention that the criterion is the conversational form of adults, and in cases where the child has an Isfahani accent, it is considered with an accent. 
13. Inflectional morphemes are counted as part of the word, not separately. For example, “Guys” was calculated as one word.
 14. “And” should be counted as a part of the word when it is attached to it, for example: Eat and. If it is produced separately, it will be compared with its own pronounced form (note that if it is produced by a child separately, it will be calculated only once, and duplicates will be removed)
 15. Rule 14 is also true for “ra,” for example, /pahasha/ (a at the end of the word was calculated as part of the word) in case of separate production, /ra/ is calculated only once. If the child used /ro-ra/ in a word in separate productions and both had the same meaning, that word was counted only once. For example: “/dar ro/ or /dar ra/.” 
16. Words like /Bozbozi/, which had a repeating component, were removed. 
17. The words that the child repeats after the examiner and actually the child says after modeling were removed. 
18. When the child produces words with different forms, such as /Shangul/, /Shangule/, and /Shangula/, were calculated, but if the child had produced more valuable and different words, the words were considered more diverse. 
19. Consider the words produced by the child as a whole as one word. For example, /Agha Gorge/, /Khanom Bozi/, and /Maman Bozi/ (of course, this rule is true when the child has produced it as a component during the entire sample). 
20. For Words that contain a meaning but are produced in different forms, calculate the most common form of use; if they are repeated equally, consider the last production of that word in 51 words. For example, /Bachehash/ or /Bachash/. 
21. Words such as /Kiyeh Kiyeh/ and /Manem Manem/, which contain two repeated components, count only one of them as a word. 
22. Do not calculate the sentence “/Ba leng dar mizane/ or /balangar mizane/.”