Dataset Card for SinFoS

Dataset Summary

The dataset contains a figures of speech in Sinhala. The figures of speech were extracted from various resources such as books and online sources. The dataset was created to assist in the translation of Sinhala figures of speech.

Schema Description

The dataset schema comprises the following attributes, designed to facilitate multi-dimensional linguistic analysis:

• Sinhala (සිංහල): The text of the figure of speech.
• Type of FOS: Top-level taxonomy distinguishing between idioms, proverbs, and maxims.
• Literal / visual image: Literal English translation (LE).
• Corresponding FOS in English: Cross-lingual equivalent (CLE).
• What it really implies: Semantic interpretation.
• Additional Context: Annotation notes and cultural metadata.

Dataset Structure

The dataset object typically looks like this:

DatasetDict({
    train: Dataset({
        features: ['Sinhala (සිංහල)', 'Type of FOS', 'Literal / visual image', 'Corresponding FOS in English', 'What it really implies', 'Additional Context'],
        num_rows: 2344
    })
})

Linguistic Classification & Taxonomy

0. Primary Classification (Major Types)

The dataset categorizes all figures of speech into five primary linguistic types based on their function and origin in the Sinhala language.

• ප්‍රස්තාපිරුළු (Proverbs)

  • Description: Traditional sayings that express a perceived truth based on common sense or experience. These are the most complex category and are further sub-classified by theme, origin, and structure (see below).

• වාග්ව්‍යවහාර (Idioms / Idiomatic Usage)

  • Description: Expressions where the meaning is not deducible from the individual words (non-compositional). These are often short phrases used in daily conversation to add color or emphasis (e.g., "eating fire" for being angry).
  • Tag: වාග්ව්‍යවහාර

• ආප්තෝපදේශ (Maxims / Authoritative Advice)

  • Description: Didactic or moral sayings derived from sages, scholars, or religious texts. Unlike common proverbs which may be folk-based, Apthopadesha carries a weight of authoritative wisdom (Sanskrit: Apta = reliable/authoritative + Upadesha = advice).
  • Tag: ආප්තෝපදේශ

• පුද්ගලික (Personal Sayings / Idiolects)

  • Description: Expressions that originated from specific individuals or characters but became popularized in general usage. These are unique to the speech patterns or history of a specific person (real or fictional).
  • Tag: පුද්ගලික

• කියමන් (General Sayings / Quotes)

  • Description: Well-known quotes, epigrams, or witty remarks that do not necessarily fit the strict definition of a proverb or idiom but are widely cited in literature and speech.
  • Tag: කියමන්

The ප්‍රස්තාපිරුළු (Proverbs) in this dataset are classified using a multi-dimensional taxonomy based on three distinct criteria: the nature of the message, the origin of the saying, and its grammatical structure.

1. Classification by Nature of Message (message_nature)

This category defines the communicative intent or the type of wisdom conveyed by the proverb.

• උපදේශ (Advisory / Instructional)

  • Description: Proverbs that function as direct advice, maxims, or warnings.
  • Example: "ළඟ තිබෙන කපුටුමහ දුර තිබෙන මොණරමහට වඛ හොඳය."
  • Tag: උපදේශ

• විශ්වාස (Beliefs / General Truths)

  • Description: Proverbs that express generally accepted social truths (Sathyadharma Pirulu) or common beliefs, rather than direct advice. These highlight realities accepted by society.
  • Example: "ඉඟුරු දීලා මිරිස් ගත්තා වගේ."
  • Tag: විශ්වාස

2. Classification by Origin / Background (origin_source)

This category traces the etymology or the source material from which the proverb emerged.

• කතා (Story-Based)

  • Description: Proverbs that summarize a specific backstory, folktale, or event into a single sentence. Understanding these often requires prior knowledge of the underlying narrative (similar to "Sour Grapes" in English).
  • Example: "අන්ධයින් හතර දෙනා ඇතාගේ ලක්ෂණ කීවා වගෙයි".
  • Tag: කතා

• ප්‍රපංච (Phenomenon / Concept-Based)

  • Description: Proverbs based on observations of nature, environmental phenomena, or abstract concepts rather than a narrative story.
  • Example: "ඇතා ගිල්ල දිවුල් ගෙඩිය වගෙයි."
  • Tag: ප්‍රපංච

• සාහිත්‍ය (Literature-Based)

  • Description: Proverbs popularized through written literature, scholarly works, or religious texts (e.g., Tripitaka, Subhashithaya).
  • Example: "මඩේ ගිලුණු ඇතා ගොඩ ගන්ට ඊට වඩා බලසම්පන්න ගජේන්ද්‍ර යෙක් ඕනා කෙරෙයි"
  • Tag: සාහිත්‍ය

3. Classification by Grammatical Structure (structure_ending)

This category classifies the proverb based on its syntax and how the sentence ends.

• උපමා (Simile / Comparative)

  • Description: Proverbs that end with comparison particles signifying "like," "as," or "similar to."
  • Keywords: Ends in wage, men, se.
  • Example: "ඉඟුරු දීලා මිරිස් ගත්තා වගේ"
  • Tag: උපමා

• ප්‍රශ්න (Interrogative)

  • Description: Proverbs that take the form of a question.
  • Keywords: Ends in interrogative particles like da.
  • Example: "උපන් ගමන් දණ ගාන්ට පුළුවනැ?"
  • Tag: ප්‍රශ්න

• වාර්තා (Report / Hearsay)

  • Description: Proverbs that end in a reporting style, implying "it is said that" or "so they say."
  • Keywords: Ends in lu or yi (similar to the Pali evam me sutam).
  • Example: "ගිය නුවණ ඇතුන් ලවාත් අද්දවන්ට බැරිලු"
  • Tag: වාර්තා

• නිශේධ (Negative)

  • Description: Proverbs that express a negation or denial.
  • Keywords: Ends in na, naha, epa.
  • Example: "එකකුට එකෙක් දෙවැනි නෑ"
  • Tag: නිශේධ

• ප්‍රකාශන/විධාන (Command / Imperative)

  • Description: Proverbs that take the form of an order, command, or call to action.
  • Keywords: Imperative verbs like karanna (do), balanna (look).
  • Example: "ගොඩ පයයි වළ පයයි"
  • Tag: ප්‍රකාශන, විධාන