RSS Feed

Tag Archives: Latin

A living Latin project: LingQ’s 60 Mini-Stories

Would you be interested in partnering with us to translate 60 Mini-Stories into Latin? This is an open, Creative Commons project in which we have the ability to create adaptations, videos, and supporting materials without fear of infringing copyright, unlike working from textbooks. LingQ, who created the original 60 Mini-Stories in English, have made them available in the public domain and are encouraging communities to translate them into their own languages. The stories revolve around ordinary situtations, feature many repetitions of vocabulary, and are already available in 39 other languages.

Read on for how you can help!

Why speak contemporary Latin?

Did you know people speak Latin on Discord? There are not one, but several Discord groups in which everyone communicates solely in Latin. The living Latin movement is growing every year, on both text-based messaging as well as on voice chats. Covid may have put a lot of the usual Latin conventions and physical meetups on hold, but international speakers of Latin are continuing to gain momentum through online platforms. It is estimated that there are tens of thousands of active speakers of Latin (though exact statistics are hard to pin down, for a group so geographically dispersed).

mihi valdē placet. I greatly enjoy this development. I want to broaden my own vocabulary and communicative ability in Latin, and I appreciate how lively and genuine the conversations have been on the various Discord groups.

But here’s my problem – as soon as I open up to talk to real people about my own everyday life, I struggle with some basic words. How do I say ordinary things like “I gotta go to work”, “I’m bored”, “coffee”, or “I’m excited for the holidays”? These phrases are not high frequency in the classical corpus, and some of the concepts are not available to Ancient Romans.

Should we not talk about normal things in Latin?

Should we limit ourselves only to the top 1000 words found in the written corpus, and not discuss the banal and mundane things of the 21st century?

While mundane, “muggle” Latin is not the central focus of my highschool classroom, I personally want to be able to sustain a good conversation in Latin and converse spontaneously about ordinary things and generally be more competent and communicative. If this is your goal too, and you want to help others in this journey, let’s make it happen!

What resources already exist?

Several resources for Conversational Latin can help us on this project.

Traupman’s Conversational Latin for Oral Proficiency

You can find videos of Traupman’s work spoken by Luke Ranieri in this playlist, and a pdf on this website. Conversational Latin for Oral Proficiency presents model dialogues full of idiomatic expressions that pertain to everyday life. It also has a great English-Latin dictionary of useful modern words, with macrons, that will greatly assist us in making the CI stories. I’m listening to the dialogues as listening practice, but I find I need to listen on repeat to get adequate repetitions for the vocabulary to sink in. Hopefully with the 60 mini-stories we can add to the pool of resources, providing more input for everyday Latin words.

Berard’s Vita Nostra: Subsidia ad Colloquia Latina, Vol 1

Vita Nostra: Subsidia ad Colloquia Latina is a handbook for learning living Latin through themed sections. Vol 1 is written, but Vol 2 is forthcoming. Here’s a detailed review of it. I haven’t yet read it as my copy is coming in the mail, but I look forward to seeing good examples of modern words used in context.

Harrius Potter, Winnie Ille Pu, et cetera

Latin translations of modern novels provide a wealth of words for everyday situations, though they are hard to search and don’t have macrons. Harrius Potter in particular has been a popular read among the Latin community, with book 1 (Philosophi Lapis) and book 2 (Camera Secretorum) translated. Book 3 was always my favourite and I wished the Latin ones hadn’t stopped at Camera Secretorum… but I digress. If you can remember how something was expressed in Harrius Potter, we have a precedent for how we can express everyday things in full sentences and in the flow of a narrative.

Latinitium’s 4 searchable dictionaries

The Smith & Hall (1871), Lewis & Short (1849), Horae Latinae (1901), and Doederlein’s handbook of Latin synoynms (1874) are searchable together at Latinitium’s 4 dictionaries page. Macrons abound! The Smith & Hall in particulary is very helpful for things as specific as “pancake”, “gun”, and “vaccine”. However, these dictionaries do date back to the early 20th century and before, so they work best for concepts that were current in their times – Smith & Hall (1871) has “miasma” though it’s missing “microbe/microorganism”, it has “atom” but no “nuclear”, “air-balloon” but no “airplane”. (A thought comes to me that this would be the perfect dictionary for a steam-punk Latin novella. Is someone writing one?)

Other Neo-Latin dictionaries

The Neo-Latin lexicon is searchable in English to Latin, and has “airplane”. It tries to indicate correct macrons, but is not always consistent (the āero- prefix needs a macron on the a).

Wiktionary has good macrons and lots of entries in the Latin language, and of particular interest to us here are the 1,917 entries in New Latin. I haven’t yet found a way to search English-Latin in Wiktionary though.

The vatican has an official contemporary Latin lexicon, Lexicon recentis Latinitatis (1997), which contains 15,000 words. The physical copies are spread out into two volumes (A-L, M-Z) which are about $30 US each (I don’t have it though). The small version is available here, albeit in Italian-Latin, and with only 565 words. The web version shows very few macrons, and I’m not sure if the physical copies indicate quantities.

How do the 60 mini-stories work?

LingQ’s 60 mini-stories are intended to circle around basic vocabulary in the language (i.e. related to ordinary life) and feature many repetitions of core words. They are arranged in order of complexity, but they are not the same as graded readers. They are brief, only about 10 lines long, and presented in three parts: story A is the story from one perspective, story B is told from another character’s perspective, and they are then followed by questions. We will record an audio reading of each story. The stories will be uploaded on LingQ, where vocabulary tools allow readers to click words and see a definition in their L1, and create flashcards.

Here’s an example of the first story in English:

A)

Mike gets up at 6:00am every morning.
He makes breakfast and drinks a coffee.
He drives to work in his car.
His work starts at 7:30am.
Mike is a cook at a restaurant.
He makes food for hungry customers.
The customers are from many countries.
They speak many different languages.
Mike can meet many friendly people.
Mike is happy when he talks to the customers.

B)

I get up at six am every morning.
I make breakfast and drink a coffee.
I drive to work in my car.
My work starts at seven thirty am.
I am a cook at a restaurant.
I make food for hungry customers.
The customers are from many different countries.
They speak many different languages.
I can meet many friendly people.
I am happy when I talk to the customers.

Questions:

1) Mike wakes up at six am every morning. Does Mike wake up early?  Yes, Mike wakes up at 6:00am every morning.
2) Mike drinks a coffee.  Does Mike drink a tea?  No, Mike does not drink a tea, he drinks a coffee.
3) Mike drives his car to work.  Does Mike drive his car to work?  Yes, Mike drives his car to work.
4) Mike’s work starts at seven thirty am.  Does Mike’s work start at seven am?  No, Mike’s work does not start at seven am.  It starts at seven thirty am.
5) Mike is a cook at a restaurant.  Is Mike a cook?  Yes, Mike is a cook at a restaurant.
6) The customers are from many different countries.  Are the customers from one country?  No, the customers are not from one country.  They are from many different countries.
7) The customers are friendly.  Are the customers friendly?  Yes, the customers are friendly.
8) Mike feels happy when he talks to the customers.  Does Mike feel happy when he talks to the customers?  Yes, Mike feels happy when he talks to the customers.

When recording the audio for these, we should allow a brief pause after each question in the question section to give the listener an opportunity to respond (usually just enough time to say “ita, sīc est” or “minimē, est hoc aliud…”). However, the listener is not obligated to offer a response, and should feel free to listen to the supplied answer to gain more input. This is based on the hypothesis that we acquire languages primarily through input and not through production.

In addition to uploading these on LingQ, we will have the stories uploaded to Wikiversity under a CC-By-SA licence.

How can we coordinate our writing?

We will translate and edit in Google Docs, so that we can easily see the state of progress and don’t double-up on stories that have already been translated.

These public links are for viewing and commenting only, just to show our progress. If you wish to write translations, send me an email (addressed to Carla, at carlaschodde@gmail.com) and I’ll add you as an editor to the Google Docs.

You may translate as many or as few stories as you want.

You can leave comments on other people’s stories, suggesting changes, or fixing typos. Use the comment feature in Google Docs or the “suggesting” mode to enter your proofreading changes.

Just some things to consider:

Style & courtesy guidelines

We want our work to be consistent and achieve the goal of reinforcing vocabulary through many repetitions. We want it to be in good Latin, but also to express modern concepts. We want it to be easy to use for learners.

  1. We’ll write macrons (ā, ē, ī, ō, ū, ȳ) for long vowels, and distinguish consonantal u as v, and consonantal i as j.
  2. Anything typed in bold black text is a draft. If we’ve reviewed and proofread something long enough, we’ll change the text to green to indicate it is “done”. Then when the audio is recorded it’ll be in golden yellow.
  3. We have freedom to rework the stories to what is easier to say in Latin. The English story doesn’t need to exactly match the Latin.
  4. We’ll take cues from the English as to the complexity of the sentence structure. In the first 10 stories, there is only a little subordination, for example. We don’t have to totally shelter grammar (eg. early subjunctives are fine), but we should try to keep our sentences brief and clear.
  5. When different phrasings are suggested, let’s prioritise expressions that have some precedent in Traupman or Harrius Potter or other recognisable modern Latin materials for consistency, or in the classical corpus, if possible.
  6. There may be multiple good ways to express an idea (eg. autocīnētum and raeda are equally valid words for “car”, according to Traupman). In the case where both phrasings are valid, and both would be valuable, we can put one of them in the main story and the other one in a comment on the Google Doc. Later we can compile all the comments into an alternate-universe version of the story that will also be useful to learners. Bonus content in the future!
  7. We want our key vocab to have many, many repetitions. That means not mixing synonyms within a story. If a story starts with raeda for car, keep using raeda throughout the whole story. The alternate version (if we make one) can have autocīnētum throughout. (Feel free to use different synonyms between stories though, just not within.)
  8. We’re imagining speaking Latin in the here and now, not writing historical fiction set in Ancient Rome. We (modern people) use modern time and dating systems, not the Roman hours-from-dawn or days-until-the-Kalends. We wear modern clothing and not togas and tunics (usually). We drink wine unmixed. We come from multiple sovereign nation states around the globe, not from provinces of the empire. We are not localising the stories in Ancient Rome, but in the modern world.
  9. Do what you want with the English names – some Latin enthusiasts use Latinised versions of their own names when speaking Latin, some use indeclinable names (especially Hebrew names), some adopt a traditional Roman praenomen. You could turn a name like “Dustin” into “Decimus”, “Dustinus”, or “Dustin” as you fancy.

Recording audio

I’m happy to record the audio for these stories, but if you’d like to make some too, you can email me (Carla, carlaschodde@gmail.com). This will be more relevant once we have a few more of them edited and proofread. We’ll indicate which story texts have been finalised in the Google Docs by turning the text green, and if a story has already been recorded it’ll be golden yellow. You can record texts which are marked green but make sure you change the status or write a note when you’re recording something so no one accidentally doubles-up on the work.

Sharing guidelines

LingQ requires us to give credit to them wherever we post these, to provide a link to their website, and to (where space allows) mention that they have many other good input-based resources like these. We would like to thank them for their generosity in writing these stories for the public domain.

I’ll make a list of contributors on the first Google Doc so we can credit your names too.

Further ideas

Let us know if you have any further uses for these stories. If you are part of other ancient or minority language communities (such as Ancient Greek, Welsh, indigenous languages, etc.), you might want to consider making your own language versions of this. They can be made even if LingQ does not yet have language support for that language. You just need to credit LingQ for the original content and provide a link to their platform. Contact Steve Kaufmann or Zoran at LingQ for advice (you can find their emails at the bottom of this page), and his team can email you materials and answer your questions.

A complete guide to Classical Latin pronunciation: the sounds of Golden Age Latin

Posted on

I have now made a Classical Latin pronunciation series on Youtube! Check out the first three parts here:

These videos are suitable for complete beginners and advanced students alike. I speak 100% in Latin, with no English in these videos. This means I make no verbal descriptions of the sounds or comparisons to any variety of English. Instead, I focus completely on demonstrating the target sounds out loud and allowing for cycles of listen-and-repeat-after-me. There is no preamble, just straight-to-the-point practice. “Show, don’t tell.”

The three videos above teach the letters, vowels, diphthongs, consonants, and various combinations of letters. I make a quantitative distinction between long and short vowels. I follow the updated Calabrese 5-vowel system, which supercedes the vowel system from Sidney Allen’s Vox Latina as a more historically likely representation of Golden Age Latin vowels. I have not yet covered syllable length, accentuation or elision in these videos, but a follow-up that explains these syllabic features could be my video on Pronunciation Tips for Scansion.

Why I created this series

I had always assumed that there were already some very good “basic” pronunciation guides to Latin, but when I searched for one, all I came up with was:

  • People reading aloud inaccurate instructions from a textbook
  • Advanced, very accurate deep-dives into single phonemes

I have no complaints about the second category, the advanced deep dives, but they are not very beginner friendly because it is hard to get a systematic overview when all phonemes are split into separate videos.

I do, however, have serious complaints about the first category!

The instructions for pronouncing Latin commonly printed in textbooks are not completely accurate representations of Latin. Because textbooks are print resources and not audio resources, they rely on “close fit” comparisons to English sounds that don’t necessarily convey the true nature of Latin sounds. The result is that the beginner will inevitably speak with their native English accent while speaking Latin – because the instructions, delivered in English, can only reference English phonemes.

There are several key weaknesses to textbook written descriptions of Latin sounds:

  • For one thing, there are multiple varieties of English pronunciation, so instructions like “o as in pot” will be interpreted differently by a standard American and a Received Pronunciation (RP) English speaker.
  • Some textbooks reference particular dialects of English, saying for example that the “r is trilled, as in Scottish”. This instruction is totally useless for me because where I live (Australia) the Scottish accent is vanishingly rare. Almost everyone I have met on this continent who identifies as Scottish has an Aussie accent.
  • Textbooks are often outright misleading when explaining aspirated vs. unaspirated consonants. For example, they instruct students to pronounce “p as in pot” and “c as in cat”. Actually, both of those consonants correspond to the aspirated versions “ph” and “ch”, as English requires all initial p, t, c to be aspirated. You can test this yourself by holding a piece of paper to your lips and feeling it flutter with the puff of air that escapes as you say “pot” or “cat”, but not “spot” or “scatter”.
  • The instruction “ph as in uphill” and “th as in anthill” lead to students and teachers making bizarre “puh huh” and “tuh huh” sounds that don’t represent aspiration.
  • Very few beginner pronunciation tutorials even hint at the nasal final -m, and those that do tend to use verbal descriptions that are very hard to replicate independently (what does it mean to “nasalise” a vowel?). This phoneme is extremely common in Latin (think of all the -am and -um endings!). Knowing that the -m is barely pronounced but is mostly a marker for nasalisation helps explain why -m is ignored for elision; I think it is a shame that students only learn that sound exists in their final years when they are scanning poetry, and not throughout the course.
  • Many Latin teachers in these videos don’t know how to pronounce the Greek “y” and say “u” instead.
  • Many Latin teachers in these videos make “ū” into something that sounds like the Greek “y” – saying “ūnus” like “oonus”.

I hope that these videos fill the gap and provide a more Latin-through-Latin approach to explaining Latin pronunciation.

Key choices

Some people may be surprised that I don’t group “ui”, “eu” and “ei” on an equal level with the other diphthongs “ae, au, oe”. I did this because there are also words in which the combinations “ui”, “eu” and “ei” are treated as separate syllables. There are only about four words with “ui” as a diphthong, and three words with “ei” as a diphthong, so I deal with these diphthongs separately and show the specific words in which they do apply.

I also decided on the Koine Greek “zz” pronunciation of “z”, which may be suprising as others have chosen Attic “zd”. The reason is simple: the Attic pronunciation of zeta as “zd” had already evolved into Koine “zz” centuries before Golden Age Latin. It is completely anachronistic to mix 500 B.C. Attic pronunciation with 100 B.C.-100 A.D. Classical Latin – the time difference is huge. The Romans of the Golden Age would have been exposed to varieties of Koine pronunciation, not Attic. To recommend “zd” for zeta in the time of Augustus would be as weird as recommending 1500 A.D. medieval French phonology for French quotations inside a 20th century English novel. At least, in my opinion. I just don’t think it is likely a contemporary of Cicero ever heard “zd”, as it wouldn’t have been normal for the Greek speakers of the period.

My vices

On recording and editing these videos, I noticed that I have a tendency to shorten final -ī into a short -i, as in when I said “vēnī, vīdī, vīcī” or “Annō Dominī”. mea culpa! Those final -ī sounds should have been long.

Let me know if you hear any other vices in my pronunciation, and I will endeavour to work on those in my future videos.

Grammar Analysis scores don’t correlate with unseen translation ability

Posted on

Some of the types of assessments Latin teachers have been traditionally setting are quite weird, the kind of things that aren’t normally done in any other languages. Consequently, it is hard to find recent research on whether what we are testing really matters.

Perhaps one of the strangest things we do is Grammar Analysis, where a student is asked to remember the “ablative of this” and the “genitive of that”.

A year or so ago, there was an alarm among Victorian Latin teachers when (due to an unfortunate miscommunication from the writers of the curriculum) we thought that Grammar Analysis would be completely removed from the new Year 12 Latin study design. At the time, we reacted quite defensively. We attended a consultation meeting with the curriculum committee where we went around the table repeating all the same things: that we needed Grammar Analysis in some form in the year 12 study design, because otherwise we would lose the rigour of the subject, students wouldn’t bother to learn grammar at earlier year levels, and the quality of unseen translation would drop.

It was a drama and half, and I shouldn’t say more about the episode, except that we got what we asked for and Grammar Analysis (“questions on accidence and syntax”) was prominently reinstated into the Year 12 study design.

But, was Grammar Analysis really the right hill to die on?

The new Victorian Latin study design gives leeway for how deeply teachers want to teach Grammar Analysis, as it is only internally assessed. How much should we emphasise Grammar Analysis among our students? How useful is it to be able to identify a subjective vs. objective genitive, or a dative of person judging, or a circumstantial participle?

For better or worse, the skill we care most about in Victorian Latin is the unseen translation. It remains just under half the marks of the final year 12 exam and is one of the best discriminators of student performance in the subject.

I decided to look into the assessment data of my year 10-12 students in the years 2018-2021 to see what metrics are the closest predictors of success in unseen translation.

Limitations

Some limitations with my study are that correlation does not equal causation – I expect at least some of these metrics will align to unseen translation without actually helping translation, because relative levels of diligence and motivation would cause the same students to do well in all hard assessments. If I plotted how the students did in Maths, it would probably correlate somewhat to how they did in unseen translation. That wouldn’t necessarily prove that skills in Maths help students translate Latin.

What this study might show, however, is a relative lack of correlation. If we expect a type of test to legitimately test student’s language skills, then we expect it to correlate with their ability to translate unseens. If we see a scattershot, perhaps we are not really measuring language skills that students use in the unseens.

A further limitation is the size of my sample. The number of data points in comparing performance within the same year was 41 (eg. John in year 10 vs. John in year 10, John in Year 11 vs. John in year 11), and the number of data points for comparing performance from a previous year to a following year was 24 (eg. John in year 10 vs. John in year 11, John in year 11 vs. John in year 12). Only 17 unique students made up the three cohorts graduating year 12 in 2019, 2020, and 2021, and there was some attrition as students moved schools during this time.

However, a positive of this data set is that all the students had the same Latin teacher the whole time (myself), and the way the assessments were taught and marked and their relative difficulty was consistent year on year.

With all this, let’s see what test scores appear to correlate most strongly to high scores in unseens.

Previous performance on unseen translation

Multiple R = 0.7243

In the graph above, I plotted the average unseen translation scores for the same student in two consecutive years. The horizontal axis represents average unseen scores in one year, and the vertical axis represents how well they did the following year.

This is a bit of an obvious point, but being previously good at unseen translations is a really good predictor for how good you’ll be at unseen translation in the future.

The Multiple R value for this correlation is 0.7243 – the tightest correlation of all the data in my study. (a value of 0 would indicate no correlation, a value of 1 would indicate strong correlation.)

Other types of assessment

The main other types of assessment (excluding seen translation) we typically do in high school Latin include:

  • Grammar Analysis
  • Vocabulary tests
  • Oral assessments
  • English to Latin translation

Grammar Analysis

In Grammar Analysis, students identify accidence and syntax in seen or prepared texts. Questions can include:

  • What case is this noun, and why was this case used? (eg. Posessive Genitive)
  • What is the person, number, tense, voice, and mood of this verb?
  • What use of the subjunctive is exemplified by this word?
  • What part of speech is this word?
  • What word does this adjective agree with?
Multiple R = 0.4039

Multiple R = 0.3646

In the first graph, I plotted how well students did in Grammar Analysis (GA) versus their unseen translation scores in the same year. In the second, I plotted their previous GA score versus their performance in unseens in the following year.

Both graphs exhibit a very wide scatter. A student in the same year could get 100% in GA but less than 85% in unseens, or conversely, they could get 90% in unseens but 60% in GA. (Multiple R = 0.4039)

Moreover, the correlation between doing well in GA one year and then doing well in unseens the next year is even weaker. (Multiple R = 0.3646)

GA does not appear to be a good predictor of success in unseen translation, especially in future years.

Anecdotally, I think all Latin teachers have experienced students who show conflicting results in GA and unseen translation. When a student comes along with great instincts at reading Latin in context but does rather woefully at remembering the Ablative of Something, we worry that all the missing information about grammar categories is going to trip them up in translation in the future, so we might advise them to work on that weakness. Conversely, when a student is doing brilliantly at GA but can’t apply the rules in translation, we might have advised them to re-read their grammar notes on such-and-such that they missed in the unseen, because they seem to respond well to explicit grammar instruction when they do GA. I’m not entirely convinced that these approaches are helpful.

If Grammar Analysis is so vital to good unseen skills, it is not clearly indicated as such by my student data.

Vocabulary

In this study, I only included vocabulary testing where it was in the traditional format of “provide one definition for this word”. The direction is always Latin-to-English. At year 10 level, I use the 20-word chapter vocab lists that come with my textbook series, but my vocabulary tests at year 11-12 level draw from very long lists of Latin words in order of frequency. For example, 100 word test is drawing from the top 100 words in Latin, the 200 word test is from the top 200, etc.

Students in these cohorts have learned lists up to 500 words long.

The typical method for learning the vocabulary is through online flashcards such as Education Perfect or Quizlet, where students write or select single English meanings for each Latin item.

Multiple R = 0.4971
Multiple R = 0.4399

Single-item vocabulary tests like these have only a slightly stronger correlation to unseen translation performance than GA scores. The scatter is very wide, especially at the lower end of performance.

I was expecting vocabulary testing to correlate strongly with “diligence” or “motivation” and thus be at least somewhat closely correlated to unseen translation skills, but it seems that being able to do well on a vocab test is not a great predictor of unseen translation performance.

In my experience, when students encounter their vocab words in a real sentence, they have trouble making the right decisions as to what meaning that word should have in its context. I tend to be pretty lenient on vocab tests and allow them to just state one meaning of the word, but in an unseen translation, some meanings are definitely more appropriate than others. An additional difficulty is that students see vocab words in “dictionary format” (with all their principal parts) in vocab tests, but need to be able to recognise them in inflected forms in an unseen. When I experimented with giving vocab tests with inflected forms of the words students were learning on the lists (eg. cēpit instead of capiō, capere, cēpī, captum), vocab scores dropped by about 50% compared to traditional tests.

Having a large vocabulary is extremely important to being able to fluently read any piece of writing. However, single-definition traditional vocab tests do not very accurately capture a student’s ability to recognise vocabulary in a translation.

I am currently looking into better ways of encouraging the development of vocabulary that responds well in context. Some of this includes experimenting with picture vocab lists where students must answer a question out loud in Latin which contains the target vocabulary feature.

Oral

My oral assessment involves getting the student to read aloud a passage of Latin, with attention to accuracy of pronunciation as well as dramatic expression. I mark students on a rubric containing five criteria: consonants, vowels, accentuation, fluency, and expression. During Covid years (2020-2021), I required students to create voice recordings, but during pre-Covid years, I listened to them perform individually.

The oral assessments were essentially just testing pronunciation, with a bit of performance flair thrown in. It was not a communicative task in the sense of simulating a live conversation in Latin.

Students were given oral assessments in years 10 and 11, but not in year 12.

Multiple R = 0.3788
Multiple R = 0.6114

Scores in oral assessment were not good predictors of student achievement in unseen translation within the same year, but intriguingly, they were somewhat correlated with unseen translation in the following year. (Multiple R = 0.6114)

Why might a student who did well on an oral one year not get the benefit of that oral in the same year but have a lag-time when they would do better on translations the next year?

I’m speaking speculatively, but it would make sense that good pronunciation might have a positive effect on contextual vocabulary acquisition.

If a student worked hard to improve their pronunciation before an oral assessment, in theory they would spend the rest of the year reading Latin with a clearer sound assigned to the words. In addition, the effort of making their performance more dramatic would involve thinking carefully about how words should be naturally grouped together into phrases, which would help in later recognising phrase boundaries in readings.

The cumulative effect of having better pronunciation could be improvements in reading that eventually improve their unseen translation abilities.

I say all this knowing that my oral assessments are quite limited in scope and my data is still pretty fuzzy, but it makes me stop and think about the significance of oral assessments if they have more of an impact on the next year’s translation ability than last year’s vocab tests and GA scores.

English to Latin translation (E2L)

Students are presented sentences in English which they must then accurately translate into Latin. The most common errors involve adjective agreement, but often there are little rules like “this verb takes the dative not the accusative” that continually pop up. Ambiguity in English is often a source of angst – should you use “suus” or “eius” for “his”? It is also brutal if students haven’t completely internalised all the types of clauses, such as which ones use the subjunctive and which ones use accusative and infinitive. This is a high-stress assessment to pretty much all students.

Students were tested on E2L in years 10 and 11, but not in year 12.

Multiple R = 0.6522
Multiple R = 0.4901

English to Latin translation (E2L) is legitimately a very challenging task. Unlike a vocab test or a grammar analysis, simple rote-learning will not cut it here. This test cannot be brute-forced with memorisation: similar to unseen translation, it requires students to really, really know their stuff.

Interestingly, E2L scores are relatively strongly correlated to unseen translation within the same year (Multiple R = 0.6522) but not as strongly correlated to the unseen skills for the following year (Multiple R = 0.4901).

I can speculate as to why the E2L didn’t have as strong a correlation to the following year’s unseen translation progress. The last E2L test my students ever face is in the first semester of Year 11, and it typically excludes some final pieces of challenging grammar such as conditionals and gerundives. While studying for their final E2L, they work hard on consolidating previously learned grammar structures such as the uses of the subjunctive, and the accusative and infinitive phrases. This is very useful to the unseen translations in year 11, which are typically a bit easier than the year 12 unseens. As unseens get filled with harder grammar not covered by their final E2L test, the correlation between how they did back then and how they do with the last pieces of grammar becomes weaker.

That may be one possibility among many. I don’t quite know what is going on, because it could also be possible that E2L is a really good measure of the same qualities (diligence, motivation) that produce students with good unseen skills. It could also be a measure of their internalisation of the Latin language, or conversely their ablity to consciously apply learned grammar rules.

Ranking the correlations

These are the test scores in order of most correlated to least correlated to performance on unseen translations:

Multiple R
Unseen Previous vs. Next year0.724301
E2L same year0.652156
Oral previous year0.611353
Vocab same year0.497062
E2L previous year0.4901
Vocab previous year0.439875
GA same year0.403941
Oral same year0.378783
GA previous year0.364568

Grammar Analysis ranks very low in this list, suggesting that it is not drawing from the same skills or characteristics that make a student good at translating unseens.

The most highly correlated factor was how well students did on unseen translation in the previous year.

The second most highly correlated factor was how well students performed in English to Latin translation in the same year.

The third most highly correlated factor was how well students performed in an oral performance of Latin in the previous year – their pronunciation and ability to add dramatic expression to a passage of Latin.

Conclusions

Greater attention needs to be given to whether we are testing the right things, and whether the tasks we set are really helpful to improving our students’ ability to read Latin in context.

I don’t like how much pressure and stress English-To-Latin translation puts on students, and it is no longer a requirement in the new study design for year 11; however, it seems to have been measuring something that really correlated closely to student performance in unseens, at least within the same year. This something could have been the students’ internalisation of the language, or it could have been their ability to monitor grammar-rules, or it could even have been their diligence and motivation.

Oral performance and pronunciation could play a much larger role in improving unseen translation skills than Victorian Latin teachers have given credit. The oral assessment is no longer a requirement in the new study design for year 11, but most teachers I have spoken to had already treated it as a very minor task or not even a formal assessment. The benefits of good pronunciation, if my limited data is showing something, are most observed in the longer term as students become better at acquiring language features. If this is the case, then we should especially emphasise oral skills and pronunciation in the early years so that the process of internalising Latin and gaining reading skills is enhanced throughout the course of study.

Grammar Analysis has very little correlation with students’ ability to translate unseens, both in the same year and in future years. It is one of the least (if not the least) legitimate language test in terms of predicting ability in unseens.

Similarly, but not quite as bad, are traditional vocabulary tests (“provide one definition for the Latin word”). Even if these are very useful words taken from large, frequency-ordered lists, doing well in a vocab test is not a good predictor for how well students can recognise the words in real contexts.

‘Palatina Medea’ or ‘Medea Palatina’? A preference for adjective-noun word order in Latin

Posted on

We’ve been told that adjectives in Latin ‘tend to’ or ‘prefer to’ follow the nouns they describe.

But on the contrary, the statistical evidence shows that Caesar and Cicero actually preferred putting adjectives before nouns.

We didn’t learn that ‘noun then adjective’ rule from reading unadapted Latin. We didn’t discover it from real usage. We learned it from our textbooks and from word of mouth. ‘They’ said that it was true and ‘they’ must right!

So when Cicero dramatically calls Clodia a ‘Medea of the Palatine’, we think he should naturally put his adjective after the noun, but he doesn’t:

507px-George_Romney_-_Lady_Hamilton_as_Medea

George Romney’s painting of ‘Lady Hamilton as Medea’. Source: [link]

Sic enim, iudices, reperietis … hanc Palatinam Medeam migrationemque hanc adulescenti causam sive malorum omnium sive potius sermonum fuisse.

For indeed, judges, you will find out… that this Palatine Medea and this migration of hers [to Caelius’s neighbourhood] has been the cause of either all the ills or, rather, all the rumours for the young man. 

However, examples like this don’t change our minds about adjective-noun order. Since they told us that adjectives ‘usually follow’ the noun, that must mean that Cicero was violating the ‘natural word order’ for dramatic effect.

It doesn’t matter how many anecdotal counter-examples we read when actually encountering the Latin; ‘this was the rule’ when we learned Latin, therefore whatever we see later confirms it. Whenever an author puts a noun first, it’s because of the rule; and whenever an author puts an adjective first, it’s because he does it for effect; ‘the exception proves the rule’ (!!).

So let’s get into the actual statistics.

Continue reading →

The accent of words ending in -que

TL;DR: Latin words ending in -que should be accented on the syllable before -que only if that syllable is (or has become) heavy; otherwise, the word should retain its original accent. If this sounds new to you, that’s probably because you’ve been following Allen & Greenough and other nineteenth century scholarship.

SPQR Roman inscription augustus imperator

The rules of accent in Classical Latin are usually very simple. Almost all words follow the formula of the ‘Penultimate Law’, which states that the accent in multiple syllable words falls on the second-last (the penultimate) syllable if this is of heavy quantity, and otherwise on the third-last (the antepenultimate).[1]

But the enclitic –que (and the other enclitics, –ve, –ne, –ce) complicates these rules.

Continue reading →

Latin tutoring: Practice literary and context questions for Aeneid VI

The golden bough, ticket to the underworld in Aeneid VI. I painted this in my Year 12 - check it out at my deviantArt gallery

The golden bough, ticket to the underworld in Aeneid VI.
I painted this in my Year 12 – check it out at my deviantArt gallery

I’ve been captivated once again by the wonderful style and substance of Vergil’s Aeneid.

But this year I’ve been particularly nerding out because my three Year 12 Latin tutoring students are all studying book VI, the journey to the Underworld, which was the book I studied when I was in Year 12.

Continue reading →

Twelve tenses: When English outdoes Ancient Greek in precision

Posted on

I have sometimes heard people say, “Ancient Greek is the most precise language in the world.” This usually comes from people who have not studied Greek for themselves and haven’t really seen its quirks first-hand.

DSC00951_2

I don’t know how best to respond. True, there are distinctions which Greek makes that English doesn’t make, but in turn there are distinctions English makes which Greek doesn’t make. (For example, “I said” and “they said” would both be expressed εἶπον [eipon] in Ancient Greek, since the first person singular and third person plural look identical in certain tenses.) As long as the idea of overall precision is left undefined, it’s not really possible to measure whether or to what extent one language is “more precise” than another.

Continue reading →

A Latin counting song and thoughts on Primary Latin

Posted on

I’m sorry that I missed an update in April, but there has been at least one reason for that. I’ve been involved in a pilot program to introduce Latin to a public primary school, and so far it has been a blast.

Here’s a song I sung to the children yesterday on my guitar.

First we learned the numbers one to ten, and then played a game. I say “unus”, and someone else says “duo” and so on in the sequence, but if two people say the next number at the same time, we have to start again at “unus”. It was fun and a good way to get them to participate in saying the numbers aloud.

Continue reading →

Latin and Greek for your pets

Posted on

Your dog can learn Latin and Ancient Greek! And everyone in the family can pick up a bit of Latin or Ancient Greek along with you and your four-legged pal.

The internet may be chocked full of cat and dog videos, but did you know there aren’t any videos of dogs responding to (grammatically correct) Latin or Ancient Greek commands? You, sir or madam, can fix that. Teach your pet Latin or Ancient Greek and film the results – I want to see your adorable furry companions nailing the ancient languages like a boss.

To get you started on teaching your animals the languages of Plato and Cicero, I’ve put together a list of suggested commands – imperatives that you can teach your dog (or cat, or rat, or any animal that can be trained to respond to verbal commands). I’ve also made a couple videos of me saying these commands in Latin and Ancient Greek, using Classical reconstructed pronunciation for both languages. Sadly, I don’t have a pet myself, but I do have Rufus here.

For the forms of the imperative, suggested commands, and some extra words that I couldn’t illustrate with my puppet dog Rufus (he can’t roll over… it really tangles him), check out the words below.

Continue reading →

Grammar or reading: which type of Latin or Greek textbook is better?

If ever you read Amazon reviews of Latin and Ancient Greek textbooks, you’ll find some very lively discussions on the relative merits of grammar- and readings-based textbooks. (If ‘lively’ is the right word to use!)

In this video, I outline the main differences between these two kinds of textbooks, and weigh in on the pros and cons of each.

In my experience, both types of textbook have complementary advantages – grammar textbooks let you advance faster, but readings textbooks give you more time to reinforce reading proficiency. What kinds of textbooks did you learn from? Which did you prefer, and why?