Creation of a list of homonyms in Brazilian Portuguese for linguistic processing tasks

Purpose: This paper presents procedures and steps to create a list of homonyms, their meanings and representative images. Methods: One hundred and three homonyms in Brazilian Portuguese were secluded through searches in text books and dictionaries and the meanings were verified. The homonyms were analyzed and selected according pre-established criteria. Twenty-six homonyms with two meanings were selected and one image was used to represent each meaning. To test the instrument, twenty elementary school children in 4th (n = 10) and 5ht (n = 10) grades were selected from a private school in São Paulo. To identify the meanings of major and minor occurrence, the students gave their understanding of the words presented to them orally. The students were also asked to name the fifty-two images to determine the degree of recognition. Percentages of accuracy were calculated. Results: Among the 26 homonyms, two were excluded because the students were unable to name them. Two images were little recognized and, therefore, replaced with others that better represented the homonyms. Conclusion: A list of 24 homonyms and 48 images (one for each meaning) was developed. The results of the presentation of these items to elementary school children in the 4th and 5th grades proved the adequacy of the list. The material is seen as efficient for use in recognition tasks and can be applied in semantic priming tasks.


INTRODUCTION
During the act of communication, different linguistic, cognitive, metacognitive and pragmatic skills are needed to capture the information and ideas contained in the sentences of oral or written discourse.Efficient access to the meaning of a word and syntactic knowledge are among the skills that function in an automatic manner, whereas self-monitoring and inferences function on the metacognitive level (1,2) .These skills are important to comprehension, as they assist in the construction and integration of a coherent model of meaning.Rapid access to the lexicon is needed for the efficient identification of the meaning of a word, with faster access leading to a better expression of the knowledge of the word as well as the automatic retrieval its phonology and meaning.Vocabulary is acquired over the course of repeated exposure, which facilitates the learning of new words (quantity of representations) and through opportunities regarding the refinement of its meanings (quality of representations) (3,4) .
In tasks involving homonyms, it is necessary to access the mental lexicon of the different meanings of a word and the discursive context must be considered to identify the correct meaning.Failure to activate relevant information and inhibit irrelevant information may be one of the factors that hinder comprehension, which could cause problems when inferences depend strongly on prior knowledge (5) .
Thus, comprehending a text, whether written or told orally, requires the individual recognition of words, the precise retrieval of their meanings and the determination of the relevance of these meanings to the discursive context so that inappropriate meanings are disregarded (2,3,6) .
All linguistic processes that involve the meaning of a word in a text are important to comprehension.Homonymy is the property of certain words to have different meanings and even different grammatical functions, despite having a single phonological form; for example: a sound mind and body; the sound of music; you sound like you have a cold (7)(8)(9) .To be understood, one must consider the discursive context in constructions with homonyms, since it is precisely this context that allows one to establish the appropriate meaning (10)(11)(12) .Studies on the identification or discrimination of homonyms generally seek the meaning with the greatest number of occurrences, which is considered the dominant meaning, and that which occurs less frequently is denominated the secondary meaning (11,12) .
Semantic priming tasks that use homonyms are excellent strategies for determining the velocity of access to the integration and inhibition of meanings.Priming is identified by the improvement in the performance of perceptive or cognitive skills related to a baseline produced in the context of previous experiences.Semantic priming regards an improvement in the velocity or accuracy of the response to a semantic cue (13) .When a homonym is presented in an isolated fashion, all meanings of the word are accessed.However, when the homonym is presented at the end of a sentence, the correct meaning is identified based on the discursive context (11,(14)(15)(16) .
Two experiments that used semantic priming tasks with homonyms proved useful with regard to evidencing three comprehension processes: accessing, integrating and inhibiting word meaning (11) .The experiments investigated these processes on the word and sentence levels, seeking to identify speed and precision in reaching meanings of greater occurrence (dominant) or less frequent occurrence (secondary) in homonyms.The results of the experiments discriminated individuals with good fluency and accurate reading skills from those with difficulties regarding the rapid access to or inhibition of a meaning of less frequent occurrence, characterizing such individuals as having poor reading comprehension (11) .Such results justify the use of homonyms to investigate reading comprehension skills with regard to isolated vocabulary words and as well as on the textual level in which the discursive context is specified.
Thus, the decision was made to study information processing for textual comprehension using semantic priming tasks with homonyms based on the belief that such tasks can demonstrate differences in the performance of schoolchildren with good and poor reading comprehension and indicate possible causes of the difficulty in comprehension related to language processing.To meet these goals, it was necessary to make a list of homonymous nouns and distinguish meanings of greater occurrence from those of less frequent occurrence using the vocabulary of schoolchildren in the 4 th and 5 th grades of elementary school.This age group was chosen based on the fact that textual comprehension is a curricular objective in these grades (17) .
The aim of this study is to present the procedures of the creation of a list of homonyms of Brazilian Portuguese, the identification of their meanings and the use of representative images of each meaning.This paper therefore describes the pilot of a study conducted to investigate the effects of semantic priming with the use of homonyms on reading comprehension.

METHODS
A pilot, prospective, quantitative, experimental study was conducted after approval from the Human Research Ethics Committee of Universidade Federal de São Paulo (UNIFESP) under process number 33431014.9.0000.5505.Signed written authorization from the director of the school was obtained.The legal guardians of the 4 th and 5 th grade students included in the study signed a statement of informed consent.The children also agreed to participate by signing a child statement of informed consent.

Selection of homonyms
To ensure that the list of homonyms was part of the vocabulary of 4 th and 5 th grade students in elementary school, a survey was performed of text books, in which 103 homonyms were identified.The meanings of these homonyms were verified by entries in two school dictionaries: Moderno dicionário da língua portuguesa [Modern Dictionary of the Portuguese Language] (18) and Minidicionário contemporâneo da língua portuguesa [Contemporary Mini-Dictionary of the Portuguese Language] (19) .
From this list of 103 homonyms, a team composed of speech-language therapists and a linguist selected those that met the following criteria: noun; elicitation of only two meanings; and words that could be represented by images.Source: Moderno dicionário da língua portuguesa (18) and Minidicionário da língua portuguesa (19) 4/12

Selection of images
Sites and files available on the internet were searched for four images that represented each of the meanings of the 26 homonyms.The speech-language therapists and linguist involved in the creation of the material met to select what they judged by consensus to be the best images to represent each of the meanings of the homonyms.

Adjustment of list of homonyms and images to target age group
To test the instrument and determine its applicability to children in the 4 th and 5 th grades of elementary school (50% of each gender; mean age = 119.7 months) and reach decisions regarding the need for changes and adaptations, a pilot study was conducted with 10 children in the 4 th grade and 10 in the 5 th grade of a private school in the city of São Paulo (Brazil) at the end of the first semester of 2015.Schoolchildren who met the following inclusion criteria were selected for the study; absence of complaints related to or indications of hearing impairment; uncorrected visual impairment; neurological, behavioral or cognitive disorders; enrollment in 4 th or 5 th grade of elementary school; and no history of having to repeat a school year.

Classification of meanings
To identify which of the meanings of each homonym was more or less frequent, the children were asked: "What meaning do you know for the words I am going to say?" Thus, the student was instructed to say what each word meant.The evocation of more than one meaning was expected to be spontaneous and enable the identification of meanings of greater or less occurrence for each word (11) .
Two responses were possible: a) a single meaning attributed; and b) two meanings spontaneously attributed to the homonym presented.When a single meaning was attributed, the response received a score of 3 points.When two meanings were attributed, the first received a score of 2 points and the second received as score of 1 point.
The classification of homonyms was confirmed based on occurrence, as follows: greater occurrence -when one meaning was attributed more than the other by at least 70% of the children, demonstrating clear dominance; less occurrence -when one meaning was attributed between 5 and 30% of the time in relation to the other meaning (11) .
In some cases, neither meaning demonstrated clear dominance (attributed 70% of the time) and two lists of homonyms were created: one clearly demonstrating which meaning was attributed more and which was attributed less and another list with the homonyms that were present in the students responses despite not demonstrating a clear dominance of one of the meanings.

Recognition of images representing each meaning of homonyms
After the 20 participants had given the meaning or meanings for each word, they were shown 52 pictures (one for each of the two meanings of the 26 homonyms) individually in a random fashion on PowerPoint and asked to name the images.Images named correctly by at least 13 students (65%) were classified as good; those named correctly by seven to 12 students were given a medium score and those named correctly by six students or less were classified as poor.Any image with a medium or poor score was reviewed by the team for the decision to maintain or exclude it based on whether the naming error occurred due to an unawareness of the meaning or the non-recognition of the image.The homonym was excluded in the former case and the image was replaced in the latter case with a different image that better represented the meaning of the homonym.For this choice, the entire image selection process was repeated.

Statistical method
The schoolchildren's answers were scored and expressed as percentage values for presentation in tables.

RESULTS
Table 1 displays the scores of the schoolchildren's answers and the percentages for each meaning of the homonyms.Meanings for the homonyms casa (button hole) and planta (blueprint) did not elicit evocations and were therefore excluded from the list.The meanings of 13 homonyms demonstrated clear dominance based on the percentage of greater occurrence.Eleven others did not achieve the expected 70%, but the two meanings were attributed by the schoolchildren and these homonyms remained on the final list.
Chart 2 displays the two lists of homonyms based on the dominance found in the meanings.
Chart 3 displays the scores regarding the correct naming of the images for each meaning of the homonyms.Based on the analysis of the data, the final list comprised 24 homonyms (13 with clear dominance and 11 without clear dominance).Appendix A displays the images of the meanings with greater or less occurrence.

DISCUSSION
Understanding comprehension in different processes enables researchers and clinicians to identify the skills and difficulties children may have.Tasks with homonyms allow the investigation of aspects of information processing, the determination of accessing, integrating and inhibiting meanings during comprehension and the identification of possible comprehension difficulties (11,20) .
The present study describes the process of selecting linguistic material and images for the construction of an instrument for use in semantic priming tasks involving homonyms that can help determine processes of accessing and inhibiting meanings during reading comprehension.
Homonymy is a phonic manifestation of a linguistic sign that has two or more meanings.Homonyms and images that represent the different meanings of these words are elements of two different languages -one verbal, comprising the word itself (linguistic sign), and the other non-verbal, comprising images, which are visual (iconic) signs that represent objects in the real world (21) .
The analysis of the occurrence of meanings attributed to the homonyms used in the present study enabled discriminating meanings of greater or less occurrence.As 75% of the schoolchildren reported two meanings for the homonyms presented (eight children in the 4 th grade and seven in the 5 th grade), the decision was made to score each answer, which permitted the identification of two groups of homonyms based on the dominance of the different meanings present in the children's answers (Chart 2).This procedure led to a final list of 24 homonyms.
In the experiments that served as the basis for the present study (11) , line drawings were used to represent the homonyms.However, the decision was made to use images closer to reality (iconic signs) in this investigation to have a better representation of each meaning of the homonyms (21) .Thus, photographs and realistic drawings were selected to represent each meaning of the homonyms iconically.
In order for the homonyms to be represented by images, those that came closest to the object in terms of representation were chosen.Thus, homonyms that presented changes in grammatical class were discarded and only nouns were used.As a noun is a class of word that names an entity, it can be easily represented by a sign of an iconic nature, thereby enabling a fast, immediate association with efficient access to its meaning (7)(8)(9) .
During the naming task, some of the images had less than 12 correct responses and were reviewed by the team to decide whether or not the figure needed to be exchanged for another.The figures for the meanings of the following homonyms were classified as good: carteira [school desk]; concha [ladle]; macaco [jack (for changing a tire)]; manga [sleeve]; pata [paw]; pilha [pile]; prato [cymbal]; tanque [tank (military vehicle)]; and terra [earth (planet)].The study demonstrated that these images adequately represented the meaning, although some of the schoolchildren were unable to name them correctly during the experiment.The failure to identify the secondary meaning may be explained by four conditioners: 1) the low frequency of the word in the daily life of the student, despite using the object (60% of the students called the ladle a "spoon" and 45% called the school desk a "chair"); 2) naming the entire object rather than the intended portion (50% of the children called the sleeve a "shirt" and 75% called the pile "books"); 3) the low frequency of the word and little or no knowledge of the object presented (30% of the students referred to the jack by its function of "lifting a car" and 15% reported not knowing the word; 65% of the students named the cymbal incorrectly ("tambourine", "drum", "to crash", "I don't know"); and 4) associations among the word, object and semantic field: 60% of the students used the word pato [duck] for pata [paw]; 60% used words other than tank (military vehicle), such as "cannon", "car of war" and "tractor"; 30% called the planet earth by the name "planet" and another 30% used the word "world".
The homonyms casa and planta were excluded from the list when the meanings attributed by the students were analyzed.Both casa [button hole] and planta [blueprint] achieved a small number of correct responses.The image for casa had only a 5% rate of correct responses, whereas the other students called the image "button", "pants" or "hole".For the image of planta, 10% of the students gave the correct response and the others used the words "house", "apartment" and "map".
The results demonstrated the need to find a different image for the homonym sombra [eye shadow], which 80% of the students named either "eye" or "makeup", and for the homonym salto [leap], which 50% of the students named "jumping", "athlete" or "sport" (Chart 3).
As expected, the results led to a list of 24 homonyms of greater or less dominance of the pairs of meanings (and corresponding images) based on the occurrence of the answers of the 20 schoolchildren (11,12) .
After the necessary changes, the homonyms and their image representations were recognized and named by the students of the 4 th and 5 th grades of elementary school, which demonstrated the suitability of the selection performed by the team.The choice of the 4 th and 5 th grades was based on the fact that the children in this age group had learned to read and write and were expected to have a vocabulary in line with the narrative and descriptive aspects that the homonyms on the list presented (17) .
The final list of 24 homonyms and the images of the meanings of greater and less occurrence were used in the priming task experiments with the isolated words as well as the words placed at the end of sentences.Homonyms that preceded the image of the meaning of greater occurrence facilitated access to that meaning, demonstrating that both materials (the list of homonyms and images) enabled the students efficient access to the meaning of each homonym (20) .
Therefore, the procedures and criteria used in the present study for access to meanings of greater or less occurrence were efficient.The images used to represent the homonyms selected were also recognized by the schoolchildren and enabled the appropriate application of the priming tasks.
Despite these results, one must bear in mind that the items have not yet been analyzed from the psychometric standpoint.Moreover, the sample that participated in the linguistic experiment and determination of the list was composed of schoolchildren in the same age group with roughly the same level of schooling (4 th and 5 th grades) enrolled in a single school in the private system of the city of São Paulo.These characteristics suggest the possibility of encountering different results in other age groups, without invalidating the list of homonyms selected, which was restricted to nouns with only two meanings.This condition is fundamental for use in experiments designed to identify or discriminate meanings of homonyms with greater or less occurrence.

CONCLUSIONS
The list created in the present study resulted in 24 homonyms and corresponding images of meanings of greater or less occurrence.The presentation of the items to schoolchildren in the 4 th and 5 th grades of elementary school demonstrated the adequacy of the homonyms and images selected.The material proved suitable for the recognition of the schoolchildren and indicated that both the homonyms and corresponding images can be used for evaluations in this age group with regard to semantic priming tasks employed to investigate the access to and inhibition of semantic information in comprehension tasks or other linguistic processing tasks that involve homonyms in Brazilian Portuguese.
Carteira-small holder made of leather or other material in which money, ID and credit cards are stored[wallet] -chair with support for writing [school desk] Casa -construction designed for habitation [house] -opening in clothing to insert buttons [button hole] Concha -calcareous hull of certain mollusks [shell] -type of large spoon for serving soup [ladle] Folha -organ that develops on the stem or branches of a plant [leaf] -piece of paper in square or rectangular shape [sheet] Galo -male poultry with crest and short, broad wings [rooster] -swelling on forehead or head resulting from a collision [bump] Laranja -citrus fruit [orange] -color, blend of red and yellow [orange] Linha -thin string used for sewing [thread] -graphic representation of a single dimension that can be considered the movement from a point [line] Macaco -mammal of the primate order [monkey] -device used to lift heavy weights [jack] Manga -part of clothing that covers the arm [sleeve] -fruit of the mango tree [mango] Pasta -portion of solid material made with any type of powder (flour, cement, etc.) and mixed or kneaded with other ingredients [paste] -type of flat leather or plastic bag/container used to store or transport documents, money, etc. [briefcase/document holder] Pata -foot or hand of an animal [paw] -female palmiped [duck] Pilha -group or mount of objects placed with some on top of others [pile] -device that emits electrical current [battery] -device connected to computer with keys that enables inputting data and initiating programs [keyboard] Terra -planet of the solar system [earth] -soil [earth] Vela -cylindrical object with fatty, combustible substance with a wick throughout the length of its center used for lighting [candle] -strong fabric unfurled from the mast of a watercraft to enable wind propulsion [sail] Xadrez -board game with 64 squares on which 32 pieces are moved [chess] -print on cloth with colors arranges in alternating squares [plaid]

Table 1 .
Sum of children's scores for each meaning of homonyms according to schooling