Introduction and context

This research will investigate whether the teaching of prose composition to a group of students with no or very limited experience helps improve their linguistic knowledge and performance. The investigation covers two different strategies: a step method and an annotation method, The new WJEC Eduqas GCSE Latin paper was first examined in 2018 and contains an optional prose composition section. This comprises three sentences to be translated from English into Latin.

The school is a non-selective girls' school located in the centre of a small town in the Home Counties. The school is rated as outstanding with very high levels of attainment and below average numbers of students with English as an Additional Language or satisfying Pupil Premium criteria (Ofsted 2013). In 2015 GCSE students achieved an A*-C pass rate of 90% compared to the national average of 69% (school website, 2015; UK Government, 2015). All students in Year 7 are taught 12 lessons of Latin and Ancient Civilisation. In Year 8 there are four classes of Latin. In Year 9 Latin again proves a popular choice with two classes running. Students are therefore interested in Latin and want to continue learning it throughout the school, with GCSE uptake in Year 10 still high.

For my project I chose to research a Year 9 class that had little experience of prose composition but would need to develop this skill for the new GCSE. The Year 9 class comprises 19 students of mixed prior attainment. Within the class there is one student who is eligible for pupil premium and there are no students identified as having any social, emotional or mental health needs or a disability of any kind. No students were identified with English as an additional language.

The whole class was taught for all three lessons which included strategies for prose composition; however, a focus group of six students was chosen for further evaluation. These were divided into three groups of two – lower, average and higher ability. Current working levels for the students ranged from 5A to 6A and so the difference in ability was not very wide. All students' identities have been removed and they will therefore be referred to as Students 1-6 (S1–6). Some evaluation of the full class' results in the upcoming exam will be analysed and compared with the parallel class' results. The other Latin class had a wider spread of ability with current working levels ranging from 5A to 7C.

Prose composition in GCSE language exams

In past Modern Foreign Language (MFL) GCSEs there was no requirement for students to translate any passages of the target language into English. For example, the AQA Spanish GCSE gave 30% of the exam to a written piece of Spanish that students composed freely. A title or topic was provided and guidance given for what kinds of information could be included (AQA, 2016).

The 2018 Spanish GCSE has introduced the translation of an English passage into Spanish (AQA, 2016b). The translation section is awarded 12 marks out of a total of 60 and consists of seven sentences arranged in a passage. The writing translations require students to both ‘convey key messages’ and to apply ‘grammatical knowledge of language and structures’ (AQA, 2016c).

All Modern Foreign Languages (MFL) GCSE exam boards have introduced translation into their exams from 2018. WJEC includes a passage or sentences in English to be translated into French depending on the tier of entry (WJEC, 2016). There is a larger amount of prose composition in MFL exams than there is in the WJEC Eduqas Latin exam.

Information on prose composition in MFL books mostly focuses on higher level writing skills than those which are expected in the new Latin exam boards' content. I decided to research primary MFL materials and see whether there was any useful information contained within them. A chapter by Alison Hurrell in a teaching MFL book aimed at primary school students gave some helpful hints in basic prose composition. These included writing ‘post-storytelling activities’ with word banks providing the different parts of a sentence; nouns, verbs, prepositions, etc. (Reference Hurrell, Driscoll and Frost1999, p.67). Students were encouraged to build sentences using a different word from each pile. In this way their sentences were formed using the correct spelling and grammar before any attempt was made to write without any prompts. This seems a particularly useful way to start thinking about composing sentences in Latin as students can practise certain skills such as identifying the different parts of a sentence and putting them into the right order.

Latin Prose Composition

The idea of Latin prose composition seems to bring up an image of students chanting verb and noun tables in a classroom followed by verbal composition into Latin and a teacher critiquing the work. Indeed the history of prose composition is bound up with the studying of Latin as compulsory up until the 1960s where Latin was a prerequisite for entry to Oxford and Cambridge universities. The Latin O level was criticised as being unsuitable with too much focus on grammar and translation which was ‘devoid of context’ (Tristram, Reference Tristram and Morwood2003, p.6). However the same can be said of the current GCSE prose composition exercises which feature three unrelated sentences without any context.

Much of the literature on Latin prose composition resigns itself to the debate of its usefulness for students. There is a fierce debate on both sides of the argument and authors of articles can be found vehemently defending their respective corners. Those articles that do provide information on prose composition tend to be focused on higher level students who are composing Latin in the style of an author, particularly Cicero. An example of this is David Matz' research which focused on students each translating a different copy of Cicero which would then be discussed in class and a vote taken to see whose version was the most Ciceronian (Reference Matz1986, pp.352–4). This exercise was used to develop pupils' understanding of Cicero's style and, although this may have been useful for higher level students, it is too complicated for Key Stage 3 students.

Wilding stated that one of the main objectives of education was for students to learn how to express themselves clearly and he felt that there was ‘no better way of reaching this goal than by studying Latin’ (1955 in Gay, Reference Gay and Morwood2003, p.77). In his approach to learning Latin Wilding included English to Latin translations which ran alongside the Latin to English translations. By using this approach Wilding stressed the importance of prose composition in learning Latin as demanding ‘mental training’ (Gay, Reference Gay and Morwood2003, p.78). English to Latin exercises also feature in the Oxford Latin Course (OLC) textbooks but these do not run parallel with Latin to English translations and have only a small part in the exercises section.

The Cambridge Latin Course (CLC) provides an inductive learning route for students focusing on reading and understanding the language. It is the most popular Latin textbook with 82% of schools currently using it (Hunt, Reference Hunt2016, p.146). There are no composition exercises for English to Latin translation. The books' aim is for students to learn about the language and culture of the Roman people. Wilding also claims this aim for his book but he chooses to do so through translation of passages into both English and Latin (Gay, p.80). The number of Latin textbooks that include prose composition indicates that it is still deemed important by some authors.

Gruber-Miller asserts that prose composition is valuable for four reasons; it helps students' communication skills, it ‘reinforces the basic ‘code of the language'’, it improves reading skills and it ‘provides an opportunity for implementing the Standards for Classical Language Learning’ (Reference Gruber-Miller2006, p.191). Gruber-Miller's belief is that communicating in Latin increases students' motivation (Reference Gruber-Miller2006, p.192). The opportunity to write stories from a specific character viewpoint can help develop cultural understanding and the writing of informative, descriptive or explanative pieces for another person to read provides further motivation. However Gruber-Miller does acknowledge that there can be negative feelings associated with prose composition. He believes that these are associated with the style of teaching and feedback to include the continual correcting of mistakes and the re-writing of sentences (Reference Gruber-Miller2006, p.190).

A different strategy found in teaching prose composition is to use famous contemporary or well-known speeches, poems or stories to translate into Latin. This is something that Kristin Lord has introduced into her university teaching but believes that it can also apply to school-age students. Lord has developed her own compositions for translating first-year Latin and makes use of quotations from books, actors and rock stars to appeal to a younger audience (Lord, Reference Lord2006, p. 9). In her compilations she ensured that composition was kept to one sentence at a time and new vocabulary was kept to a minimum. This seems to be useful advice for Year 9 prose composition.

Saunders (Reference Saunders1993) firmly believes that prose composition is a rewardable learning experience that can be undertaken by students of all abilities as long as it is taught well and at a suitable level of length and difficulty. In order to do this at a beginners level, such as my class will be doing, there is a need to reinforce grammar, syntax, vocabulary and word order (Saunders, Reference Saunders1993, p. 392). Both Lord and Saunders insist on making beginners' prose composition manageable in terms of length and difficulty.

Davisson (Reference Davisson2000) uses familiar vocabulary when setting prose composition and bases her exercises on sentences previously translated in class with some small changes made, such as changing noun number or different vocabulary. Davisson then gets students to write a paragraph in English but with a major grammatical error in each sentence that a Roman might have made (Davisson, Reference Davisson2000, p. 76). Students have to explain correctly the ‘rule of Latin grammar or idiom’ and the English grammar point. Examples were the exclusion of the definite article or placement of a verb after the direct object in English. By making these mistakes students are supposed to develop their understanding of Latin grammar and its use in sentences. As Davisson's (Reference Davisson2000) title suggests her recommendations are suited to intermediate Latin students, equivalent to higher-end GCSE or AS level. However, this strategy might work on a simpler level once students have had some experience of English to Latin translation.

A further exercise that Davisson (Reference Davisson2000) uses is the re-telling of a Latin passage from another character's perspective. She asks students to write a paragraph and provides details of the grammar she expects, such as third declension nouns, so that students are required to choose more complicated vocabulary. The passage then goes through several drafts as Davisson asks students to include more ‘sophisticated constructions’ and complicated grammar such as subjunctives and participles (Davisson, Reference Davisson2000, p.79). Students can write in the style of a particular Roman author and use literary techniques such as anaphora or word order to improve their writing. Davisson marks the different stages in students' work on the passage as a continual assignment so that students can see their progress. I think this strategy would work really well for an AS class but it could also be simplified for a Key Stage 3 class who were writing a simple story on gladiators or Pompeii for example.

Beneker's (Reference Beneker2006) research was inspired by Davisson's but focused on first-year Latin students. He felt that a different preparatory phase was necessary which was specifically tailored to beginning Latin students. His idea was to focus on ‘analysing English sentences and translating them into Latin’ in order to help students ‘develop their analytical skills through composition’ (Beneker, Reference Beneker2006, p.2). The approach begins with a formulaic method where each English word is identified and labelled. A sentence is displayed on the board and students analyse each word. Beneker then encourages students not to look words up but to try out vocabulary to see if it works in the sentence and discuss with classmates. His belief is that by making mistakes in translation, such as the addition of esse in translating the imperfect verb ‘were conquering’, and discussing why it is wrong, students will gain a better understanding of word choice and style (Beneker, Reference Beneker2006, p.3). Students label each word in the sentence with its function and its form in Latin. They then add the Latin vocabulary to each word and write the sentence out. Students finally rearrange the sentence to provide a more fluent Latin style. This approach seems to be suitable for beginning Latin students and follows a systematic formula for translating into Latin. It also provides a useful revision of key grammar terms and the purpose they have in a sentence.. Beneker makes some criticisms of his research: the students often get confused if sentences are too complex; they skip steps causing more mistakes to be made; and the first prose composition exercises can be daunting (Beneker, Reference Beneker2006, pp.6–7). His research is small-scale and a work in progress but it offers good suggestions for breaking down prose composition into manageable parts and labelling words before systematically putting them back together in a sentence.

Eales' (2015) study of students' confidence in prose composition focused on indirect statements but has some use for studying student motivation. He found that students felt less confident in English to Latin translations because their knowledge of grammar was weaker than other skills in Latin such as reading and translating from Latin. (Eales, Reference Eales2015, p.24). His study found that students considered syntax to be the most difficult aspect followed by vocabulary, recognising constructions and word order (Eales, Reference Eales2015, p.24). Students' actual difficulties were mostly centred on forming the correct tenses for verbs. Although Eales' (2015) study is only based on five students it does supports Beneker's (Reference Beneker2006) and Saunders' (1993) research that students can become easily confused by lots of grammar, whereas a simplistic formula aids successful prose composition.

Research Questions

The review of literature and my knowledge of the students focused my research on three questions. These required an analysis of current student understanding of prose composition and introducing strategies that might help to improve comprehension. The research questions decided upon were

1. 1. What do students seem to have difficulty with when writing simple Latin sentences? (RQ1)

2. 2. What do students say they have problems with when writing Latin sentences? (RQ2)

3. 3. How did students use the strategies I used to develop their understanding of prose composition? (RQ3)

Teaching Sequence

Three lessons were available to carry out the research with the opportunity to collect further data from an exam in late April. Each lesson was an hour long. There was no extra staff available for the lessons so the research methods chosen needed to be carried out by me solely. During the first lesson students were observed by me and their comments were noted, by me, on how they translated the sentences. Observation took place during all three lessons and copies of work were taken. Questionnaires were produced for the first lesson to assess how confident the students were with English to Latin translation and which aspects of Latin they felt least and most confident in.

Due to an exam scheduled for the week after the Easter holidays the lessons featuring prose composition needed to include some aspect of revision. This informed the teaching that I would use in the lessons. I wanted to ensure that the students were applying knowledge of grammar to their prose composition work so that they were able to revise. The other Year 9 class would have a minimum of two lessons focused on revision and I did not want my class to be disadvantaged.

The research conducted comprised of observations, questionnaires, a quiz and students' work. The first lesson in my teaching sequence was used to generate data on students' English to Latin translation. This took the form of a prose composition exercise consisting of three sentences to translate. Students were given the vocabulary but no other help was provided. A questionnaire was also used to assess students' opinions of Latin with regard to current areas of interest and understanding plus their thoughts on the translation were sought before and after the task.

The next two lessons were designed to teach the students strategies to help with prose composition. These took the forms of grammar consolidation and revision and practising constructing sentences. At the end of the second taught lesson the students were given three sentences to translate which were to be compared with their first attempt.

After the three lessons I was informed that the Year 9 Latin exam was to include an optional English to Latin translation and therefore a further opportunity to collect data on students' translating skills was possible. Students' answers from the paper were therefore also analysed.

Data and Findings

In line with Hunt's statement that direct translation rather than free composition becomes an exercise in solely testing knowledge of vocabulary and grammatical structures, vocabulary retention was not assessed (Hunt, Reference Hunt2016, p.37). In order to increase the motivation of students and to test specific teaching methods rather than vocabulary retention, all of the exercises with the exception of the initial exercise, contained the necessary vocabulary.

Students were selected for the research after all three lessons had been taught due to the changing attendance during lessons. Students were selected on the basis of their ability, with two students picked within each ability level. Table 1 shows the assessment data for the selected students.

Table 1.

| Student data

Students were not told how well they had performed on any of the exercises so that their motivation levels were not affected by a perceived or actual low score or result. The correct sentences were not discussed with the students at any stage and only translation exercises that were worked through together in class were ever analysed.

In order to assess students' initial ability in prose composition they were asked to translate three sentences from English into Latin.

What do students say they have problems with when writing Latin sentences? (RQ2)

Before the first translation exercise students were asked to complete a short questionnaire to provide information on their likes and dislikes in Latin lessons to see if these had any correlation with their prose composition results. Their answers are displayed in the following table 2.

Table 2.

| Aspects enjoyed most in Latin and the easiest and hardest areas for the students

Students stated they enjoyed taking part in Kahoot! quizzes the most. The easiest aspect of Latin was felt to be the translating of stories in the CLC books. All students felt that grammar, either all of it or particular aspects, was the hardest to learn. Those students identified as lower ability were more specific on the grammar points they did not know.

Before students actually translated the sentences they were asked to look at them and answer a few questions on the perceived ease or difficulty they might face. The summary of their answers is below in Table 3 and provides answers to Research Question 2.

Table 3.

| Perceptions of translation and confidence levels

During the translation exercise an observation of the conversations between students was carried out and some of their dialogue was noted. The students were advised that they could discuss with each other how best to translate the sentences. Below is a sample of the conversation.

• S1: If I have stuff it's ok (Referring to books/iPads to help translate). Walking is ambulat, so more than one - nt.

• S5: How do you spell that? i makes it plural

• S6: festinat begins with festin. How do you change the ending?

• S1: How do you tell it's imperfect? I'm forgetting everything - vocab. It's annoying I know all the words.

• S5: into is just in

• S4: pulchrum. What is hurry?

• S1: iratus. It's a girl.

• S5: Wow! Would you end puella?

• S1: Carrying is like portat.

• S5: How do you put it in the past tense?

• S4: If it is a girl I'm gonna do an a.

• S1: portat. I think it's just portat.

• S3: What is into? in, oh yeah.

What do students seem to have difficulty with when writing simple Latin sentences? (RQ1)

Common problems identified in the exercise were translating present tense verbs in the imperfect, not using the accusative case for both nouns and adjectives where required, not making iratus agree with puella and adding in a form of the verb esse when not required. These difficulties with the translation provide the answers to Research Question 1. In order to concentrate the intervention on three specific areas I decided not to include adjectives in future exercises. Students were given the vocabulary if they were unsure of any spellings but were only provided with the nominative singular for nouns and the present first person for verbs. The correct spelling of words was not taken into account when assessing the sentences although endings of nouns and verbs were. Five of the students attempted all three sentences, one student completed one sentence fully and one partially.

Things that students were generally successful in in their translations were putting servus into the nominative plural and translating the imperfect. Some students used the correct word order. Students' perceived difficulties and those aspects they said they found easier are summarised below in Table 4.

Table 4.

| Student perceptions of easiest and hardest sentences.

The second lesson in the teaching sequence took place two days later. In order to understand more fully the nature of the grammar that students said they found hardest I asked the whole class to carry out a quiz. The questions included in the quiz had been put to a Year 8 class during the previous week and had provided a good indication of issues they faced with grammar. The Year 9 class seemed fairly confident in answering the questions and the majority got the answers right. A few students struggled to remember the terminology for nouns relating to subject and object but were very clear that accusative nouns ended with a m. Not all the questions were answered as the final two were more challenging and only some students answered them. However, those that did could explain how the imperfect and perfect were formed in English.

As students had had difficulties with translating verbs into the correct tense I used part of the lesson to revise present, imperfect and perfect tense verbs. Students were asked to identify verbs in the three cases. There were still some problems with translating into the present tense. The particular issue was with the inclusion of ‘am’ and ‘ing’ in the sentence ‘I am walking’. Students were identifying verbs ending in ‘-ing’ as being imperfect rather than present tense. It was necessary to explain the different ways in which the present tense could be translated. Students were informed that this could be either written as ‘I walk’ or ‘I am walking’. The use of the imperfect was also revised so that students could see the difference between the two tenses in English. Students found identification of both the present and imperfect tense in Latin easier than in English. Students were also taught that there was no need to add in an extra word for ‘am’ or ‘was’ in Latin.

There was also some revision of nouns in the nominative and accusative, both singular and plural. Students were asked to highlight in different colours nouns and verbs in a sentence to ensure that they could identify them correctly. Students found identifying nouns by both number and case easier than the work on verbs. The exercises set during the lesson were in line with Saunders' findings, whereby the work was of a suitable length and difficulty (1993, p.392). As with Lord's recommendations vocabulary was kept to a minimum using words that students were familiar with (2006, p.9).

The first strategy I used with the students was to label all the words in the sentences provided. Five sentences containing fairly easy vocabulary from Stage 8 of the CLC Book 1 were used. The first two sentences were displayed on the white board so that all the students could see how to annotate each word. Students differed in their approach to labelling each word. Student S2 highlighted verbs in one colour and nouns in another but did not annotate them. Student S4 labelled the nouns fully but only added the tense of the verb. The answers to the sentences were discussed in class so that all students could understand how they had been translated.

The final lesson was used to show students another strategy to help with translating from English into Latin which was based on Benecker's research (2006). This technique involved students following a five-step approach to writing prose composition. The main idea was for students to write down the vocabulary for each word and then make changes to the word in order to write the correct Latin.

How did students use the strategies I used to develop their understanding of prose composition? (RQ3)

Finally, students were asked to translate sentences using both methods, annotating each word and the step method. Vocabulary sheets with all the words used in the lessons were handed out so that the vocabulary could be accessed easily. The students made a concerted effort on both methods and completed the translations. A questionnaire was given to the students to analyse which method they preferred and the results are summarised below.

All students except one said they felt more confident when translating from English into Latin than they did previously. Their original rating for confidence was not relayed back to them and so the scores for confidence levels were as unbiased as possible. The majority of students scored their confidence as either the same or higher than before the intervention showing a small positive change. The findings from the questionnaire also show that most students preferred the step method when doing prose composition. It is interesting to note that the student who did not report any change in her confidence level after the intervention preferred the annotation method; this was in opposition to all the other students.

The grammar that students found they were more confident using in translation was variable, as was their least confident aspect. This shows that although confidence had mostly increased students were still finding particular things more challenging than others, with students S1, S2 and S6 naming the same aspect of grammar before and after intervention.

Student S2 struggled with plural nominative nouns and perfect tense verbs but was able to translate all the present tense verbs correctly. Word order was completely correct for all sentences using the step method; annotating did not seem to help with this. Verb tenses for all the step method sentences were also all correct. For this particular student the step method seemed to be more useful and results show an improvement in her prose composition skills. However, there are still mistakes in the sentences mostly with noun number and case. Student S2's perceived difficulties with prose composition were ‘tenses and endings’. The student remained having problems with plural noun endings and neither strategy for translating them helped. This was one of the perceived weaknesses by the student and her analysis seems correct here. The student's other perceived weakness was forming verbs correctly in all three tenses. However, the step method seemed to provide a better student outcome regarding verb tenses, as all verbs were correctly translated in all three tenses. Student S2 had actually seemed to improve her translation of verbs in different tenses. However, it must be remembered that this was a small-scale study and data was analysed after a recent intervention. Results gathered in future exercises may not show the same level of improvement as techniques are forgotten or students lapse into the same translation mistakes.

Other students also made some improvements in their prose composition but the caveats above should be taken into consideration. Both students S1 and S4 made only one mistake in their translation using the step method. Both were able to compose all the words, except one, correctly using the step method. This was to not put villa into an accusative form in sentence three. Both students wrote all sentences using the correct word order, with all verbs positioned at the end of the sentence. The step method enabled both students to compose English to Latin translations well and to overcome their perceived difficulties as noted in Table 3.

In the annotation method, students S1 and S4 made the same mistake in not making villa accusative, even though S4 had labelled it correctly as a singular, accusative noun. However S1 did not label the words as fully as S4 and only included the labels ‘noun’ and ‘verb’. She also did not use the correct word order and wrote the sentences in the same order as the English, leaving the verb in the middle of the sentence. For student S1 it seems that the step method helped to write more accurately in Latin.

Student S6 felt that her weakest area in prose composition was word order. This was borne out by her translation using the annotation method where the verbs were placed in the same position as the English. However, when using the step method S6 placed all the verbs at the end of the sentence correctly. The step method for this student also seemed preferable. This is supported by the student's confusion during the annotation exercise where verbs were labelled as ‘nominative’. Her present tense verb was written as lacrimat although the subject of the sentence was plural. The step method provided sentences with fewer mistakes and the present tense verb was correctly written as a plural. This student struggled with putting nouns into the accusative when using both strategies. This was not a part of translation that she identified as being a weakness or felt less confident in but is one that seems to be shared by many of the students.

Table 5.

| Student confidence before and after using strategies and the preferred strategy

Table 6.

| Grammar that students perceived to be more confident with, after the intervention

Table 7.

| Analysis of Prose Composition Exam Questions

Student 5 was the only student that thought the intervention had made no difference to her prose composition skills and was the only student to prefer the annotation method. However, the results show that her sentence structure was better using the step method as the correct word order was used with verbs placed at the end of the sentence. In the annotation method verbs were positioned in the middle of the sentence where a direct object was included. As with other students S5 did not translate direct objects into accusative nouns accurately in either method.

Student S3's results are only partially analysed as she had to leave for a music lesson and did not complete the step method translation. However, her prose composition using annotation displayed similar results to the other students. The word order was incorrect with verbs not placed at the end of the sentence. Verb tenses and conjugations were correct. Student S3 also had problems with translating the direct object into the accusative case but her perception of nouns in Table 3 was that they were easy as they did not need changing. Her belief that changing words from singular to plural was the most difficult area of translation did not seem to be evident in her prose composition.

Four weeks after the intervention, the Year 9 classes sat a Latin exam which featured an optional prose composition section which contained three sentences. All of my class answered all the questions fully. In the parallel Year 9 Latin class, five students did not attempt the section and three answered it only partially. Hopefully this shows that my intervention did give my class confidence in attempting prose composition. This could be supported by the choice of five of my class not completing the alternative Latin to English translation. The exam results of the six sampled students is summarised below which includes problems identified in the translation.