LLM Indexing

Prompt

I want to work on improving the index of the book whose root is at source/book/learning_machines.tex. Start by reading indexing.txt, which describes best practices for indexing. Then look at the final six pages of source/book/learning_machines.pdf, which contains the book’s index. Then create a list of issues with the index and proposed solutions. This list should be in indexing_updates.md. Possible issues include: entries with too many entries (solution could include selecting only a few key entries, changing individual pages to page ranges, breaking the item into sub-items nested under the main item, removing the term all together), similar entries which should be grouped together or nested, and other issues named in indexing.txt. Look at the source text (in source/book/chapters/) when diagnosing problems and proposing solutions.

Plan

Issues identified by comparing the printed index (pp. 349–353) against the ASI Best Practices for Indexing guide and the source .tex files in source/book/chapters/. Source file evidence is given where useful.

1. Typo: Engish language learner

Current: Engish language learner, 21, 22, 62, 63, 75, 78, 82, 83, 119, 312, 313, 323–325

Issue: Missing “l” — should be “English language learner.” The source uses \index{Engish language learner} throughout.

Fix: Global find-and-replace in all chapter files: \index{Engish language learner} → \index{English language learner}.

Approved.

2. Metatopic entry: language!English — 358 occurrences, nearly every page

Current: – English, 4, 10, 18, 22, 25–27, 36–40, 44, 46, 48–55, 57, 61–63, 66–68, 71–73, 75, 76, 79, 80, 84, 91–98, 100, 104, 105, 108, 110, 112, 117, 119–121, 126–128, 133–135, 137, 142, 143, 148, 157, 161–163, 165, 174, 175, 192, 193, 222, 224, 227, 229, 231–233, 237, 239–242, 244–246, 250, 251, 265, 266, 275, 279, 281, 285, 287, 290, 291, 294, 298, 299, 307, 308, 311, 316, 318, 320, 322, 323, 325–327, 332, 333, 337, 339–342

Issue: With 358 source occurrences, English appears on virtually every page of the book. This is a metatopic condition (ASI §1.1–1.2): the book is entirely situated in English-dominant technological and educational contexts, so indexing every mention of English gives the reader no actionable guidance. The locator string spans nearly the whole index and will frustrate users.

Fix: Remove the language!English entry entirely, or drastically reduce it to only the pages where English is specifically analyzed or theorized (e.g., discussions of Standard English ideology, English-only policy, English as a hegemonic force). Suggested reduced list (based on introduction and chapter content): keep approximately pages 4, 14–16, 35–38, 94–97, 307–308, 337–341. Add a cross-reference: language – English. See also monolingualism; standard language ideology; English-only.

We will not use “English” as an entry at all, but will index subtopics under English. Examine all indexed uses of English in the manuscript, create subtopics, and re-assign current English entries to subtopics. For occurrences of English which do not pertain meaningfully to a subtopic, remove the indexing all together.

3. Near-metatopic entry: power — 194 occurrences

Current: power, 1, 3, 9–11, 15–22, 24, 26, 38, 63, 64, 66, 69, 79–85, 98, 100, 101, 120, 123, 129, 130, 156, 157, 166, 183, 186, 189–192, 197, 211–214, 219, 222, 223, 227, 246, 251, 255, 267, 305–310, 314–319, 321–323, 325–328, 331–333, 336, 338, 340, 343, 344

Issue: Power is one of the book’s central organizing concepts, appearing across all 11 chapters. A flat list of 50+ page references is textbook underanalysis (ASI §4.10, §5.8). Chapter 6 (Puglisi) specifically theorizes three dimensions: episodic, dispositional, and systemic power. The introduction also devotes a section to power relations.

Fix: Break into subentries. Proposed:

power
  – episodic, 189–191, 197
  – dispositional, 190
  – systemic, 186, 190, 191, 197, 211–214
  See also hierarchies; hegemony; resistance

Keep only the most essential general-reference pages (e.g., introduction’s definitional discussion: 9–11, 15–22) at the main entry level.

Agreed. Synchronize the treatment of power with the treatment of English above.

4. Ambiguous and overlong entry: code — 152 occurrences

Current: code, 2, 11–17, 20–23, 27, 41, 47, 73, 74, 94, 105, 106, 111, 116, 118, 120, 129, 131, 132, 134–139, 141, 143, 145–148, 188, 221, 227, 232, 233, 239–244, 251, 264, 315, 332, 341

Issue: “Code” is used with at least three distinct meanings across the book: (a) computer programming code; (b) “code” as a theoretical concept bridging race and computation (Ruha Benjamin’s framework, introduced on pp. 11–17); (c) zip codes and other social-classificatory codes. Collapsing these into one undifferentiated string makes the entry nearly useless (ASI §4.10). Note that New Jim Code already has its own separate entry (12, 335, 339).

Fix: Add subentries:

code
  – computer programming, 116, 118, 120, 129, 131, 132, 134–139, 141, 143, 145–148, 188, 232, 233, 239–244
  – as classificatory scheme, 2, 11–17, 20–23, 27, 94, 105, 106, 111, 221, 227, 251, 264, 315, 332, 341
  See also New Jim Code; algorithms

(The introduction’s pp. 11–17 defines both senses and is the key definitional passage.)

Agreed. Treat this the same way as English and Power above. Ensure that the three senses of code in the introduction are subtopics.

5. Misleadingly labeled entry: hacking — 103 occurrences in one chapter

Current: hacking, 13, 20, 92–94, 99–107, 109–113, 124, 264

Issue: 103 of the 116 source occurrences are in Chapter 3 (Creanga), where “hacking” refers exclusively to a coined concept: language hacking — creative translingual adaptation of smart speakers by multilingual families in Romania. The plain entry hacking suggests computer hacking/security, which is not the book’s subject. The introduction uses the term in connection with Beltrán’s (2022) framework. A reader looking up “hacking” in the traditional sense will be misled; a reader interested in the specialized concept cannot find it by its full name.

Fix: Replace the entry with language hacking (or hacking, language) and add a cross-reference hacking. See language hacking. The page range 92–113 spans the chapter body; rather than listing every page, reduce to key passages: language hacking, 13, 20, 92–94, 99–101, 109–113, 124, 264.

Approved.

6. Underanalysis: literacy — 106 occurrences

Current: literacy, 3–5, 7, 8, 23, 25, 27, 62, 80, 82, 92, 106, 112, 117, 119, 120, 123, 124, 126–128, 130, 132, 140, 156–161, 163–168, 173–178, 184, 188–190, 194, 309, 311, 314, 324, 333, 336, 340, 343, 344

Issue: 38-page locator string with no subheadings. digital literacy and algorithmic literacy are currently separate top-level entries but logically belong as subentries of literacy. The Science of Reading entry (which is about a specific literacy framework) should cross-reference here.

Fix: Add subentries and fold in related entries:

literacy, 3–5, 7, 8, 23, 25, 27, 82, 92, 112, 127, 130, 132, 333, 336
  – algorithmic, 343, 344 [currently separate entry; remove top-level]
  – digital, 62, 80, 112, 160, 164, 166, 167, 177, 184, 189, 190, 194, 309, 311 [currently separate entry; remove top-level]
  – foundational/reading, 156–161, 163–168, 173–178
  See also computational literacies; Science of Reading

Remove the standalone digital literacy and algorithmic literacy top-level entries, replacing with See literacy cross-references, or simply absorb them.

Approved.

7. Underanalysis: translanguaging — 123 occurrences, book-spanning list

Current: translanguaging, 3, 16, 26, 36, 39, 40, 43, 51, 52–96, 98, 99, 112, 113, 117, 118, 120, 121, 123, 126–132, 135, 137, 141, 147, 158, 163, 166–168, 174–178, 219, 222, 224–226, 228, 231–233, 238, 240, 244–246, 255, 307–309, 314, 327, 332–334, 336, 341

Issue: Translanguaging is a central theoretical framework applied in nearly every chapter. The flat list is unhelpful. The range 52–96 (the entire Chapter 3 span) is particularly egregious — it collapses an entire chapter’s analysis into undifferentiated pages. Note also the suspicious-looking range 52–96 which appears to be the result of consecutive page-by-page indexing rather than selective indexing.

Fix: Add subentries differentiating key uses:

translanguaging
  – theory/framework, 3, 16, 26, 103, 117–121, 255, 307–309
  – in technology design, 36, 39–43, 51–56, 307–309, 314, 327
  – as pedagogy, 126–132, 135, 141, 147, 219, 224–226, 232, 238, 314
  – and language hacking, 92–94, 99–101
  See also translingualism; critical translingual computing

Approved.

8. Duplicate entries: identity! and language! for the same terms

Current: Several terms are indexed simultaneously as both identity!X and language!X at every occurrence, producing two index subentries pointing to identical pages.

Evidence from source: In d-chapter-02-Le.tex, every mention of Hindi reads \index{language!Hindi}\index{identity!Hindi}. Same pattern in other chapters for Kurdish and Cherokee.

Term language! pages identity! pages
Hindi 40, 46, 49–51, 53 40, 46, 49–51, 53
Kurdish 265, 266 265, 266
Cherokee 276, 287, 294–296, 299, 301 276, 287, 294–296, 299, 301
Zapotec 284, 291 284, 291

Issue: Identical locators at two index locations is redundant (ASI §5.9), giving the reader no reason to prefer one entry over the other.

Fix: For each case, keep whichever categorization best fits the chapter’s focus:

  • Hindi (Chapter 2 / Vogel, interface design): context is primarily the language on multilingual interfaces → keep language!Hindi, remove identity!Hindi
  • Kurdish (Chapter 9 / Thomas, linguistic queering): context is primarily the language in translingual practice → keep language!Kurdish, remove identity!Kurdish
  • Cherokee (Chapter 9 / Thomas): the chapter explicitly addresses Cherokee as both language and cultural identity in a language revitalization context; here double-posting is justified — but the pages are still identical, so add a brief cross-reference rather than duplicating: under language!Cherokee, add See also identity!Cherokee
  • Zapotec (Chapter 9 / Thomas): same as Cherokee; retain language!Zapotec, add cross-reference from identity

Approved.

9. Double-posted Latinx — inconsistent page lists

Current:

  • Top-level: Latinx, 6, 18, 20, 43, 117, 120, 126, 160, 168, 229, 282, 283, 291, 297, 311, 342
  • Subentry: identity – Latinx, 6, 18, 20, 43, 117, 120, 126, 160, 168, 225, 228, 229, 282, 291, 297, 311, 342

Issue: Double-posting is legitimate (ASI §7.0–7.1), but the locators differ slightly (the subentry includes 225, 228; the top-level entry includes 283). Per ASI §7.4, double-posted entries must have the same locators.

Fix: Reconcile the two lists so they are identical. Review pages 225, 228, 283 in the source to determine which entry should include them, then synchronize both.

Move all Latinx to identity – Latinx

10. Entries that should be nested or consolidated

10a. online search absorbed into online search and brokering

Current:

  • online search, 62, 64, 65, 71, 72, 77, 79, 83
  • online search and brokering, 62–65, 67, 68, 74, 76–78, 80–85

Issue: The online search pages are a near-subset of online search and brokering. Both entries draw on the same chapter (Chapter 1, Salas). The narrower term adds nothing a reader could not find under the broader one.

Fix: Remove online search as a standalone entry. Optionally add it as a subentry of online search and brokering:

online search and brokering, 62–85
  – by teenagers, 63, 68–70, 74, 76–85
  – by immigrant families, 62–64, 66, 67, 77, 85

Or simply remove it and let online search and brokering serve both purposes.

10b. network analysis absorbed into social network analysis

Current:

  • network analysis, 26, 187, 202, 213
  • social network analysis, 26, 187, 197–199, 202, 213, 214

Issue: The network analysis pages (26, 187, 202, 213) are a strict subset of social network analysis (26, 187, 197–199, 202, 213, 214). Both refer to the same methodology in Chapter 6 (Puglisi).

Fix: Remove network analysis. Add a See cross-reference: network analysis. See social network analysis.

Current:

  • immigrants, 61, 67–69, 71, 117, 148, 228
  • immigrant families, 62–64, 66, 67, 77, 85, 176
  • immigrant youth, 66, 333

Fix:

immigrants, 61, 67–69, 71, 117, 148, 228
  – families, 62–64, 66, 67, 77, 85, 176
  – youth, 66, 333
  See also online search and brokering; parents; teenagers

10d. decolonization nested under decolonial perspectives

Current:

  • decolonial perspectives, 220, 228, 251, 252, 257, 258, 267, 284
  • decolonization, 252, 340

Fix: Nest: add decolonization as a subentry under decolonial perspectives, or add decolonization. See also decolonial perspectives cross-reference.

10e. Computer Science Education nested under Computer Science

Current:

  • Computer Science, 116–125, 127, 129, 146, 148, 149, 306, 335
  • Computer Science Education, 116, 117

Issue: The education subfield is discussed only within the broader Computer Science context in the same pages.

Fix:

Computer Science, 116–125, 127, 129, 146, 148, 149, 306, 335
  – education, 116, 117

10f. dual language bilingual education nested under bilingual education

Current:

  • bilingual education, 3, 16, 18, 311, 328
  • dual language bilingual education, 126, 170, 171

Fix:

bilingual education, 3, 16, 18, 311, 328
  – dual language, 126, 170, 171

10g. family storytelling absorbed into storytelling; storytelling cross-referenced with storywork

Current:

  • family storytelling, 223
  • storytelling, 221, 223, 224, 226, 231, 238, 242
  • storywork, 219–225, 227, 229–233, 238, 243–245, 333

Fix:

storytelling, 221–224, 226, 231, 238, 242
  – family, 223 [absorb family storytelling]
  See also storywork

storywork remains as a separate entry (it is a distinct Indigenous theoretical concept, not merely “storytelling”).

10h. ways of being / knowing / speaking — add cross-references

Current: Three separate entries:

  • ways of being, 36, 130, 156, 256, 268, 274, 336, 345
  • ways of knowing, 19, 20, 24, 120, 123, 167, 177, 219, 224, 244, 247, 252
  • ways of speaking, 157, 224

Issue: These three parallel terms are epistemological companions. Readers interested in one are likely interested in all three, but there are no cross-references connecting them. ways of speaking has only two pages and could be folded in.

Fix: Add See also cross-references linking ways of being ↔ ways of knowing ↔ ways of speaking. Consider folding ways of speaking (only 2 pages) under ways of being or merging with ways of knowing.

10i. technoptimism and technopessimism — merge into single entry

Current:

  • technoptimism, 2
  • technopessimism, 2

Issue: Both entries point to the same single page (p. 2), where they are introduced as a contrasted pair. Keeping them as two single-page entries wastes space and splits access to the same discussion.

Fix: Merge: technoptimism/technopessimism, 2 (or keep separately but note each with a See also cross-reference to the other).

Approved.

11. locations subentries — several minor locations warrant removal

Current: The locations entry has 22 subentries including:

  • – Arizona, 291 (1 page)
  • – Australia, 194 (1 page)
  • – France, 49, 98 (2 pages)
  • – Italy, 194 (1 page)
  • – Somalia, 68, 69 (2 pages, participant origin only)
  • – Vietnam, 75, 78 (2 pages, participant origin only)
  • – the Middle East, 126 (1 page, passing mention)

Issue: Per ASI §5.8, page lists for single- or double-page entries in an already-long subentry array add clutter without providing value. These are mostly passing mentions of participant origins rather than substantive geographic discussions.

Fix: Remove subentries with fewer than 3 pages where the location is mentioned only as participant background (Somalia, Vietnam, Arizona, Australia, Italy, the Middle East, France). Keep: California, New York City, Romania, Los Angeles, Mexico, New Jersey/Newark, Canada, Europe/Western Europe, Great Lakes, Michigan, the European Union (all with substantive discussion of 3+ pages).

Approved.

12. digital tool subentries — several tools with 1–3 page references

Current: The digital tool entry has 22 named tools as subentries. Several appear on very few pages:

  • – Amira Learning, 4 (1 page)
  • – Epic, 170, 174 (2 pages)
  • – GPS, 48, 53, 54 (3 pages)
  • – Khanmigo, 2 (1 page)
  • – Sanas AI, 338 (1 page)
  • – WhatsApp, 41, 47 (2 pages)

Issue: Tools mentioned only once or twice in passing do not meet the threshold for meaningful indexing. A reader who wants to find what the book says about Khanmigo will find only a brief mention on page 2.

Fix: Remove subentries for tools discussed on fewer than 3 pages (Amira Learning, Epic, GPS, Khanmigo, Sanas AI, WhatsApp) unless they are substantively analyzed rather than merely mentioned. For tools that are analyzed in depth — especially Scratch, Hummingbird, Google, Zoom, YouTube — retain and consider whether their long page lists warrant subentries.

Approved.

13. Missing cross-references

The following pairs of related entries have no cross-references linking them, which will frustrate readers working through related concepts:

From To (missing cross-ref)
Artificial Intelligence See also machine learning; large language models; algorithms
machine learning See also Artificial Intelligence; algorithms
translanguaging See also translingualism; critical translingual computing
translingualism See also translanguaging
racism See also anti-Blackness; raciolinguistic ideologies; racialization
raciolinguistic ideologies See also racism; standard language ideology; language ideologies
language ideologies See also raciolinguistic ideologies; standard language ideology; monolingualism
colonialism See also decolonial perspectives; resistance; imperialism
social capital See also social network analysis; network analysis
Science and technology studies Add: STS. See Science and technology studies

Approved.

14. parents and teenagers — overlong lists from single chapter

Current:

  • parents, 54, 62–64, 66–68, 70–85, 92, 156, 157, 161, 162, 164–168, 171, 172, 177, 227, 230, 245, 246, 278, 291, 324
  • teenagers, 63, 68–70, 74, 76–85, 124

Issue: The bulk of parents (pages 62–85) and nearly all of teenagers (pages 63–85) come from a single chapter (Chapter 1, Salas) which is specifically a study of immigrant parents and teenagers doing online search together. The long string of individual pages 70–85 for parents should be given as a range.

Fix: Consolidate the Chapter 1 references into ranges: parents, 54, 62–85, 92, 156, 157, 161–168, 171, 172, 177, 227, 230, 245, 246, 278, 291, 324. For teenagers, consolidate: teenagers, 63–85, 124. Add cross-references:

parents. See also immigrant families; online search and brokering
teenagers. See also immigrant youth; online search and brokering

Approved. Also examine references to parents and remove less important ones.

15. hacking entry note — also check introduction pages 13, 20

The introduction (p. 13) uses hacking in connection with Beltrán’s concept of hacking as political resistance by Latino tech workers; p. 20 uses it in a more general theoretical frame. Both these pages should appear under language hacking only if the discussion fits; otherwise these introduction references may warrant a separate subentry:

language hacking
  – as political resistance (Beltrán), 13, 20, 264
  – in smart speaker use (Creanga), 92–94, 99–101, 109–113, 124

Approved.

Summary of changes by type

Type Entries affected
Typo fix Engish language learner → English language learner
Remove or drastically reduce (metatopic/near-metatopic) language!English, power, translanguaging
Add subentries (underanalysis) code, literacy, translanguaging, power, language hacking
Rename entry hacking → language hacking
Remove duplicate entries identity!Hindi, identity!Kurdish, identity!Zapotec (identical pages to language counterparts)
Reconcile double-posted pages Latinx (top-level vs. identity!Latinx)
Nest under parent entry digital literacy, algorithmic literacy (under literacy); Computer Science Education (under Computer Science); dual language bilingual education (under bilingual education); immigrant families/youth (under immigrants); decolonization (under decolonial perspectives); family storytelling (under storytelling); network analysis (under social network analysis)
Merge technoptimism/technopessimism; online search into online search and brokering
Prune minor location subentries locations!Arizona, !Australia, !France, !Italy, !Somalia, !Vietnam, !the Middle East
Prune minor digital tool subentries digital tool!Amira Learning, !Epic, !GPS, !Khanmigo, !Sanas AI, !WhatsApp
Add cross-references 10+ pairs listed in §13