> For the complete documentation index, see [llms.txt](https://ai4commsci.gitbook.io/formosanbank/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://ai4commsci.gitbook.io/formosanbank/the-bank-architecture/corpora/montgomerytexts.md). # The Montgomery Texts ## The Montgomery Texts This corpus contains three Amis texts with word-level segmentation, English word glosses and English sentence translations, digitized from [this publication](https://github.com/FormosanBank/FormosanBank/blob/main/Corpora/MontgomeryTexts/CodeAndDocs/Original.pdf): > Tamih and Montgomery, Robert L. (1962) Amis text. Work Papers of the Summer Institute of Linguistics, University of North Dakota Session: Vol. 6, Article 6. 10.31356/silwp.vol06.06 The article uses a 1962 SIL field transcription that it never names; it was identified afterwards (see Notes). Each file's first sentence is the text's **printed title**, which the article sets over its own word-aligned English gloss line, so the titles are published as sentences like any other. The texts are segmented into words but **not into morphemes**: the hyphens in the forms mark morpheme boundaries, but the article gives no morpheme-by-morpheme gloss line, so no morpheme tier is invented. The `standard` tier is a real transliteration of the original through Akiw's letter correspondences below, not a copy of it: 185 of the 391 sentence- and word-level forms differ, including every sentence-level form. > **The `PHON` (IPA) tier was regenerated in September 2026, and 568 of its 788 values changed.** If you took a copy before then, refresh it. Three things moved: the retired bare-tilde variant notation (`o~u`) became the current `[o|u]` form; segmentation hyphens were stripped, as a PHON should not carry them; and a wrong cell in the orthography profile was corrected — it had given the letter `l` the sound /d/, where Akiw's own example (`tamlaw` → `tamdaw` "people", /tamɬaw/) shows it writes /ɬ\~ɮ/. Before the fix, the same word was transcribed /dafak/ from the original tier and /ɬafak/ from the standard. They now agree on every element, and no value carries an uncertainty marker. *** #### Corpus Statistics | |

Amis
unknown

| | ------------------------- | ---------------------- | | Word count | 358 | | Total audio | 0 | | Transcribed | 0 | | Untranscribed | 0 | | Translated words | | | English | 358 | | Mandarin | 0 | | Japanese | 0 | | Dutch | 0 | | Morphologically segmented | 0 | | Glossed words | 0 | *** #### Notes Many thanks are owed to Hsu Cheng-Wen "Akiw", who helped identify the orthography. Her notes follow. 1. Letter Correspondences These are some identified differences between the original orthography and the current standard. |Original | Ortho113 | | | ts | c | | | ? | ' | glottal stop | | ř | r | | | r | l | | | l | d | e.g., tamlaw → tamdaw (“people”) | The status of the diacritic on `f` (which is rare — one word, `ma-sasi?af̆a?ařaw`) is uncertain, and remains so. The published XML writes it as `f` with a combining breve. A high-resolution reading of the scan suggests the mark is a caron, matching the `ř` elsewhere on the same line, but a single glyph on a 1962 mimeograph is not enough evidence to change a published letter, so it stands. The reasoning is recorded in the corpus's [`CodeAndDocs/source_discrepancies.md`](https://github.com/FormosanBank/FormosanBank/blob/main/Corpora/MontgomeryTexts/CodeAndDocs/source_discrepancies.md). The import of hyphens, which are used sporadically, has not been ascertained. In the published data they are treated as morpheme boundaries: kept on the word tier, and removed from sentence-level `standard` forms so that tier reads as running text. The correspondence table audits cleanly through IPA — all five rows confirmed, no information loss. The `dialect` attribute is left as `unknown`: the article does not say, and nothing in the orthography narrows it. (Southern Amis in particular is ruled out rather than suggested — Ortho113 marks `f` and `o` as unused in Southern, and these texts are full of both.) 4. Two English equivalents for one word The article's gloss line sometimes gives two English readings separated by a slash — `good/holy`, `noon/lunch`, `trip/walk`, `whole/all`, `and/with`. The Amis word does not vary, only its English rendering, so these are not published as separate sentences. Each is a primary translation plus a second one marked `ver="alt"` on the same word. No published translation keeps a slash. 2. Omission of the central vowel “e” It appears that the central vowel is missing in many places: mahřek -> maherek ("finish") tstsay -> cecay ("one") pafri -> pafeli ("give") 3. Glides are used inconsistently. soar → sowal (“words”) tsiwro → ciwlo(“elders” / 長老— loanword ) *** ## Copyright **License:** CC BY-NC-ND 4.0 The individual authors cannot be located. The publisher — the Summer Institute of Linguistics / University of North Dakota — released the article as **CC-BY-ND**, and FormosanBank has **interpreted that licence liberally** in order to publish the texts with their glosses and translations. Publication rights were confirmed by the FormosanBank maintainer on 23 August 2026. If you are a copyright owner, please reach out to us. *** #### Citation In accordance with our [Terms of Use](/formosanbank/additional-resources/terms-of-use.md), if you use this corpus or any product derived from this corpus in any publication, you must cite both FormosanBank and: * Tamih and Montgomery, Robert L. (1962) Amis text. Work Papers of the Summer Institute of Linguistics, University of North Dakota Session: Vol. 6, Article 6. 10.31356/silwp.vol06.06