I need some help here, especially from the organizers. I noticed that the translations have some inconsistent quotations. I have seen numerous examples of translations where there is a closing double quotation without a corresponding opening double quotation. Here are two examples:
\nYou can also notice the double quotation at the end of the translation without a corresponding opening quote. Sometimes these double quotations at the end of the translation are two (\"\"), like in the following translation:
\nFor the multiple quotations at the end of the translations, is this a CSV formatting issue, or they were actually inserted by the translators? Moreover, I can't seem to find the determinatives (since it's stated that they are marked with braces) in the training set.
", "messages": [ { "id": "3379803", "postDate": "12/20/2025 18:48:39", "content": "I need some help here, especially from the organizers. I noticed that the translations have some inconsistent quotations. I have seen numerous examples of translations where there is a closing double quotation without a corresponding opening double quotation. Here are two examples:
\nYou can also notice the double quotation at the end of the translation without a corresponding opening quote. Sometimes these double quotations at the end of the translation are two (\"\"), like in the following translation:
\nFor the multiple quotations at the end of the translations, is this a CSV formatting issue, or they were actually inserted by the translators? Moreover, I can't seem to find the determinatives (since it's stated that they are marked with braces) in the training set.
", "rawMarkdown": "I need some help here, especially from the organizers. I noticed that the translations have some inconsistent quotations. I have seen numerous examples of translations where there is a closing double quotation without a corresponding opening double quotation. Here are two examples:\n- Šalim-Aššur's attorney questioned Man-mahir. The attorney said: *According to a verdict of the City the lapis lazuli of Šalim-Aššur that was deposited in the City - whether investors of Šalim-Aššur have seized it or have taken it to cover the debts of Šalim-Aššur's sons, they must return both silver and lapis lazuli. He who does not return it will have it taken away by force. When Šalim-Aššur was alive he gave you lapis lazuli for purchases. Either give me the lapis lazuli or give me the price in silver of the lapis lazuli, so I may bring it to the City. Do not create difficulties for me, the attorney! I shall indeed remember what was sworn by an oath by the City and the king - that he has obstructed me. He refused to give either the lapis lazuli or the price of the lapis.\"* This is not a formal court case, only a witnessed statement. Witnessed by Šu-Ištar son of Erradi, by Ugi... son of Ennam-Aššur.\"\n- Aššur-šamšī and Ennam-Suen seized us and Ennam-Suen said to Aššur-šamšī: *Give me the silver that accrues to me\"*. Aššur-šamšī answered: \"I will / can not give you the silver, for I have ... ... you ... ... for ... ... \". Ennam-Suen said: \"Why is it me that you are... ... ... ... ... ... me? Do it to your sons!\" Aššur-šamšī answered: \"I will personally take your expenses for my account\". In the presence of Zuba (and) of Puzur-Adad; Ali-abum, son of Zia was our colleague. For these words the Kanesh colony made us available and this tablet (with) our testimony we presented before Aššur's dagger.\"\n\nYou can also notice the double quotation at the end of the translation without a corresponding opening quote. Sometimes these double quotations at the end of the translation are two (\"\"), like in the following translation:\n- To Aršiahšu and Peruwa say, thus Kassu-bēlī: Please, my brothers my lords, 30.3333 shekels of silver and 15 minas of purified copper under my seals Bēlum-bāni is bringing you. Buy for me cleaned wool, do not try to save even half a shekel of silver! I will pray for you before Aššur, Send him quickly, it is winter time.**\"\"**\n\nFor the multiple quotations at the end of the translations, is this a CSV formatting issue, or they were actually inserted by the translators? Moreover, I can't seem to find the determinatives (since it's stated that they are marked with braces) in the training set.", "votes": null }, { "id": "3381045", "postDate": "12/23/2025 16:19:14", "content": "Thank you for your observations! You are correct to note that quotation marks have not been used \"by the book\" in all instances. Some of this stems from what I'll call \"lost in translation\". In other words, some of these translations were provided in another language and were translated into English for this competition. Other instances of this can be attributed to the scholar / editor of the text (remember, most are not native English writers) who apparently is only using the double-quotes (\") as a marker for when direct speech ends, since that is the most difficult aspect to infer from the Akkadian language.
", "rawMarkdown": "Thank you for your observations! You are correct to note that quotation marks have not been used \"by the book\" in all instances. Some of this stems from what I'll call \"lost in translation\". In other words, some of these translations were provided in another language and were translated into English for this competition. Other instances of this can be attributed to the scholar / editor of the text (remember, most are not native English writers) who apparently is only using the double-quotes (\") as a marker for when direct speech ends, since that is the most difficult aspect to infer from the Akkadian language.", "votes": null }, { "id": "3390577", "postDate": "01/13/2026 12:47:45", "content": "So, the test set will also have such issues regarding the quotation marks?
", "rawMarkdown": "So, the test set will also have such issues regarding the quotation marks?", "votes": null }, { "id": "3390679", "postDate": "01/13/2026 17:57:42", "content": "Just for clarity, it is important to note that there are no question marks in the transliteration of the test set. These were removed.
\nFor the translations, the most common use of question marks are based on question words / interrogatives in Akkadian:
\nQuestions can also be stated with the negative (lá): \"isn't it the case that…?\" / \"are you not …?\"\nSometimes translations try to infer when a statement is actually a question. This would be difficult to train.
\nEven more difficult to train (and perhaps pointless) are cases of modern insertions of unsure readings, interpretations, and translations. This can take the form of question marks in parentheticals: \"I have given a donkey (to you?), but you refused.\", as well as in the text itself: \"From Aššur-nādā? to Enlil-bāni? say the following…\" OR \"I gave you 60? shekels\". These are unfortunate and it would have been wise to remove these, but without editing the test set, they will impact everyone's score in the negative equally.
", "rawMarkdown": "Just for clarity, it is important to note that there are no question marks in the transliteration of the test set. These were removed.\n\nFor the translations, the most common use of question marks are based on question words / interrogatives in Akkadian:\n\n- kî / kê (ki / ke-e) = how?\n- mannu (ma-num / ma-nu-um) = who?\n- a(na) + mannu = amannu (a-ma-nim) = to whom?\n- mati = when?\n- matima = whenever?\n- mīnu = what? (also: mīnum, minû, mīna, mīnam, mīni, mīnim, mīnumma, mīnamma, mīnammi, mīnimma, mīnimmi, min), (e.g. mì-nam)\n- a(na) + mīnu (amīnu) = for what purpose? why?\n- miššum = why? (mì-šu-um / mì-šu)\n\nQuestions can also be stated with the negative (lá): \"isn't it the case that...?\" / \"are you not ...?\"\nSometimes translations try to infer when a statement is actually a question. This would be difficult to train.\n\nEven more difficult to train (and perhaps pointless) are cases of modern insertions of unsure readings, interpretations, and translations. This can take the form of question marks in parentheticals: \"I have given a donkey (to you?), but you refused.\", as well as in the text itself: \"From Aššur-nādā? to Enlil-bāni? say the following...\" OR \"I gave you 60? shekels\". These are unfortunate and it would have been wise to remove these, but without editing the test set, they will impact everyone's score in the negative equally.", "votes": null }, { "id": "3390682", "postDate": "01/13/2026 18:09:00", "content": "I think he means quotation marks, not question marks.
", "rawMarkdown": "I think he means quotation marks, not question marks.", "votes": null }, { "id": "3390692", "postDate": "01/13/2026 18:32:04", "content": "My question here was referring to the double-quotes and not the question marks.
", "rawMarkdown": "My question here was referring to the double-quotes and not the question marks.", "votes": null }, { "id": "3390695", "postDate": "01/13/2026 18:42:41", "content": "Sorry for the confusion. Yes this problem of only partial quotation marks persists in the test data too. It is quite common to see the opening of a quote without the closing marks. This is especially the case following \"um-ma\" (which introduces direct speech), e.g. PN replied: \"….. The reason for this is both that sometimes the translator didn't include them, but also, when dividing the sentences, the quote would sometimes fall into another sentence boundary.
", "rawMarkdown": "Sorry for the confusion. Yes this problem of only partial quotation marks persists in the test data too. It is quite common to see the opening of a quote without the closing marks. This is especially the case following \"um-ma\" (which introduces direct speech), e.g. PN replied: \"..... The reason for this is both that sometimes the translator didn't include them, but also, when dividing the sentences, the quote would sometimes fall into another sentence boundary.", "votes": null } ], "comments": [ { "id": 3381045, "author_name": "deeppast", "author_url": "", "post_date": "12/23/2025 16:19:14", "content": "Thank you for your observations! You are correct to note that quotation marks have not been used \"by the book\" in all instances. Some of this stems from what I'll call \"lost in translation\". In other words, some of these translations were provided in another language and were translated into English for this competition. Other instances of this can be attributed to the scholar / editor of the text (remember, most are not native English writers) who apparently is only using the double-quotes (\") as a marker for when direct speech ends, since that is the most difficult aspect to infer from the Akkadian language.
", "votes": null, "replies": [ { "id": 3390577, "author_name": "einherjer", "author_url": "", "post_date": "01/13/2026 12:47:45", "content": "So, the test set will also have such issues regarding the quotation marks?
", "votes": null, "replies": [ { "id": 3390679, "author_name": "deeppast", "author_url": "", "post_date": "01/13/2026 17:57:42", "content": "Just for clarity, it is important to note that there are no question marks in the transliteration of the test set. These were removed.
\nFor the translations, the most common use of question marks are based on question words / interrogatives in Akkadian:
\nQuestions can also be stated with the negative (lá): \"isn't it the case that…?\" / \"are you not …?\"\nSometimes translations try to infer when a statement is actually a question. This would be difficult to train.
\nEven more difficult to train (and perhaps pointless) are cases of modern insertions of unsure readings, interpretations, and translations. This can take the form of question marks in parentheticals: \"I have given a donkey (to you?), but you refused.\", as well as in the text itself: \"From Aššur-nādā? to Enlil-bāni? say the following…\" OR \"I gave you 60? shekels\". These are unfortunate and it would have been wise to remove these, but without editing the test set, they will impact everyone's score in the negative equally.
", "votes": null, "replies": [ { "id": 3390682, "author_name": "anaphase21", "author_url": "", "post_date": "01/13/2026 18:09:00", "content": "I think he means quotation marks, not question marks.
", "votes": null, "replies": [] }, { "id": 3390692, "author_name": "einherjer", "author_url": "", "post_date": "01/13/2026 18:32:04", "content": "My question here was referring to the double-quotes and not the question marks.
", "votes": null, "replies": [ { "id": 3390695, "author_name": "deeppast", "author_url": "", "post_date": "01/13/2026 18:42:41", "content": "Sorry for the confusion. Yes this problem of only partial quotation marks persists in the test data too. It is quite common to see the opening of a quote without the closing marks. This is especially the case following \"um-ma\" (which introduces direct speech), e.g. PN replied: \"….. The reason for this is both that sometimes the translator didn't include them, but also, when dividing the sentences, the quote would sometimes fall into another sentence boundary.
", "votes": null, "replies": [] } ] } ] } ] } ] } ], "raw_markdown_by_id": { "3379803": "I need some help here, especially from the organizers. I noticed that the translations have some inconsistent quotations. I have seen numerous examples of translations where there is a closing double quotation without a corresponding opening double quotation. Here are two examples:\n- Šalim-Aššur's attorney questioned Man-mahir. The attorney said: *According to a verdict of the City the lapis lazuli of Šalim-Aššur that was deposited in the City - whether investors of Šalim-Aššur have seized it or have taken it to cover the debts of Šalim-Aššur's sons, they must return both silver and lapis lazuli. He who does not return it will have it taken away by force. When Šalim-Aššur was alive he gave you lapis lazuli for purchases. Either give me the lapis lazuli or give me the price in silver of the lapis lazuli, so I may bring it to the City. Do not create difficulties for me, the attorney! I shall indeed remember what was sworn by an oath by the City and the king - that he has obstructed me. He refused to give either the lapis lazuli or the price of the lapis.\"* This is not a formal court case, only a witnessed statement. Witnessed by Šu-Ištar son of Erradi, by Ugi... son of Ennam-Aššur.\"\n- Aššur-šamšī and Ennam-Suen seized us and Ennam-Suen said to Aššur-šamšī: *Give me the silver that accrues to me\"*. Aššur-šamšī answered: \"I will / can not give you the silver, for I have ... ... you ... ... for ... ... \". Ennam-Suen said: \"Why is it me that you are... ... ... ... ... ... me? Do it to your sons!\" Aššur-šamšī answered: \"I will personally take your expenses for my account\". In the presence of Zuba (and) of Puzur-Adad; Ali-abum, son of Zia was our colleague. For these words the Kanesh colony made us available and this tablet (with) our testimony we presented before Aššur's dagger.\"\n\nYou can also notice the double quotation at the end of the translation without a corresponding opening quote. Sometimes these double quotations at the end of the translation are two (\"\"), like in the following translation:\n- To Aršiahšu and Peruwa say, thus Kassu-bēlī: Please, my brothers my lords, 30.3333 shekels of silver and 15 minas of purified copper under my seals Bēlum-bāni is bringing you. Buy for me cleaned wool, do not try to save even half a shekel of silver! I will pray for you before Aššur, Send him quickly, it is winter time.**\"\"**\n\nFor the multiple quotations at the end of the translations, is this a CSV formatting issue, or they were actually inserted by the translators? Moreover, I can't seem to find the determinatives (since it's stated that they are marked with braces) in the training set.", "3381045": "Thank you for your observations! You are correct to note that quotation marks have not been used \"by the book\" in all instances. Some of this stems from what I'll call \"lost in translation\". In other words, some of these translations were provided in another language and were translated into English for this competition. Other instances of this can be attributed to the scholar / editor of the text (remember, most are not native English writers) who apparently is only using the double-quotes (\") as a marker for when direct speech ends, since that is the most difficult aspect to infer from the Akkadian language.", "3390577": "So, the test set will also have such issues regarding the quotation marks?", "3390679": "Just for clarity, it is important to note that there are no question marks in the transliteration of the test set. These were removed.\n\nFor the translations, the most common use of question marks are based on question words / interrogatives in Akkadian:\n\n- kî / kê (ki / ke-e) = how?\n- mannu (ma-num / ma-nu-um) = who?\n- a(na) + mannu = amannu (a-ma-nim) = to whom?\n- mati = when?\n- matima = whenever?\n- mīnu = what? (also: mīnum, minû, mīna, mīnam, mīni, mīnim, mīnumma, mīnamma, mīnammi, mīnimma, mīnimmi, min), (e.g. mì-nam)\n- a(na) + mīnu (amīnu) = for what purpose? why?\n- miššum = why? (mì-šu-um / mì-šu)\n\nQuestions can also be stated with the negative (lá): \"isn't it the case that...?\" / \"are you not ...?\"\nSometimes translations try to infer when a statement is actually a question. This would be difficult to train.\n\nEven more difficult to train (and perhaps pointless) are cases of modern insertions of unsure readings, interpretations, and translations. This can take the form of question marks in parentheticals: \"I have given a donkey (to you?), but you refused.\", as well as in the text itself: \"From Aššur-nādā? to Enlil-bāni? say the following...\" OR \"I gave you 60? shekels\". These are unfortunate and it would have been wise to remove these, but without editing the test set, they will impact everyone's score in the negative equally.", "3390682": "I think he means quotation marks, not question marks.", "3390692": "My question here was referring to the double-quotes and not the question marks.", "3390695": "Sorry for the confusion. Yes this problem of only partial quotation marks persists in the test data too. It is quite common to see the opening of a quote without the closing marks. This is especially the case following \"um-ma\" (which introduces direct speech), e.g. PN replied: \"..... The reason for this is both that sometimes the translator didn't include them, but also, when dividing the sentences, the quote would sometimes fall into another sentence boundary." }, "source": "meta" }