Speech-sound patterns: description before a phonological rule
A rule that fits known forms has not earned every prediction. Keep sound description, conditioning environment and the evidence needed to choose between analyses separate.
Course topic: Phonetics and Phonology
Study route and exercise: University Reading List editorial project. How we curate · Sources checked:
On this page
Scope and responsibility
The open route focuses on American English and has Dutch comparisons; it is not a language-general phonology course. All paradigms here are invented, with no audio experiment or claim about an actual language.
Choose your starting point
Learn to describe sounds free
Use Broeders/Gussenhoven to distinguish articulation, transcription and spelling.
Broaden phonetic evidence
Use Ladefoged/Johnson for production, acoustic description and practical transcription.
Analyse an alternation
Use Odden after introductory linguistics and IPA to formulate competing rules.
Compare the three books
An Introduction to American English Phonetics: A Theoretical Guide to American English Pronunciation
Ton Broeders; Carlos Gussenhoven
Free English sound foundation
- Choose it when
- You need a controlled introduction to an English pronunciation model.
- Use another route when
- You need a universal accent standard or complete cross-language phonological theory.
- Selected reading
- Use the consonant, vowel and transcription discussions; keep the separate pronunciation-practice companion distinct.
- Limits and trade-offs
- A teaching accent is an analytical target, not a hierarchy of speakers or a diagnosis.
Try before choosing
Describe [k], [g], [tʃ] and [dʒ] by voicing, place and manner.
Check your work: Do not treat an affricate as a spelling sequence or collapse voiced and voiceless forms.
A Course in Phonetics
Peter Ladefoged; Keith Johnson · Edition: 7 · 2015
Broader phonetics course
- Choose it when
- You want articulatory and acoustic perspectives beyond one English variety.
- Use another route when
- You only need a compact rule-analysis workbook.
- Selected reading
- Use the selected seventh edition’s articulation, transcription and acoustic discussions, matching its contents.
- Limits and trade-offs
- Transcription and acoustic measures are evidence representations, not phonemic analyses by themselves.
Try before choosing
List what the supplied bracketed strings tell you and what a sound recording would additionally be needed to measure.
Check your work: The paper strings supply categories, not durations, spectra or measured speaker variation.
Introducing Phonology
David Odden · Edition: 2 · 2013
Phonological reasoning
- Choose it when
- You can read IPA and want to test generalizations across word forms.
- Use another route when
- You need your first account of speech organs or transcription symbols.
- Selected reading
- Read the second edition’s discussions of alternations, underlying representations and rule formulation.
- Limits and trade-offs
- A compact rule is a hypothesis constrained by data; elegance alone does not prove uniqueness.
Try before choosing
Predict the new -e forms under both supplied hypotheses before seeing the follow-up.
Check your work: Hie predicts affricates; Hi predicts stops. Do not select a winner from training fit alone.
Check your preparation
What do brackets mean here?
They mark supplied surface transcriptions. No automatic conclusion about the unique underlying representation follows.
What mathematics is required?
Only checking each root/suffix pairing against a rule. Keep the trigger set and voicing correspondence explicit.
Does a counterexample refute every possible related rule?
No. It rejects the precise deterministic candidate it contradicts. A narrower or morphologically conditioned hypothesis may remain.
A six-step study plan
Steps completed: 0 / 6
These are your own study marks, not an assessment. They are not sent to this site.
-
Describe the segments
Selected reading: Open theory text: consonants.
What to do: Make a feature table for four consonants.
Check your work: Voicing is preserved across k→tʃ and g→dʒ.
-
Organize the paradigm
Selected reading: Odden: alternations.
What to do: Arrange four roots by -a and -i forms.
Check your work: Compare forms of the same root, not unrelated words alone.
-
State two candidates
Selected reading: Odden: rules.
What to do: Write both trigger sets before checking the table.
Check your work: One is {i,e}; the other is {i}.
-
Audit training fit
Selected reading: Phonetics course with paradigm.
What to do: Mark every supplied form as predicted or contradicted.
Check your work: Both candidates fit eight forms.
-
Predict before revealing
Selected reading: Odden: testing analyses.
What to do: Record the two predictions for R1-e and R3-e.
Check your work: Predictions differ even though training fit agrees.
-
Update and limit
Selected reading: All three routes.
What to do: Evaluate the held-out forms and state an unresolved rival.
Check your work: Reject Hie, retain Hi as compatible, not uniquely established.
Worked exercise
Two rules fit the same eight forms
Invented roots R1–R4 combine with exactly the shown suffixes: R1-a [paka], R1-i [patʃi]; R2-a [tuka], R2-i [tutʃi]; R3-a [tega], R3-i [tedʒi]; R4-a [moga], R4-i [modʒi]. Candidate Hie changes stem-final k→tʃ and g→dʒ immediately before either i or e. Candidate Hi makes exactly those changes only immediately before i. Elsewhere each leaves the stop unchanged. Compare training fit and predict R1-e/R3-e.
Assumptions
Treat the surface strings, root/suffix identities and candidate operations as exact. Both candidates use the stop stems visible with -a; there are no extra changes or exceptions. This is not proof of those stems as the only possible underlying analysis.
- For every -a form, neither candidate triggers, so [k] or [g] remains. For every -i form, both trigger and preserve the supplied voicing correspondence. Both fit all eight.
- For R1-e, Hie predicts [patʃe], while Hi predicts [pake]. For R3-e, Hie predicts [tedʒe], while Hi predicts [tege].
- No -e form occurs in the training packet. The fit therefore does not determine which extension is correct. Rule choice requires discriminating evidence, not another restatement of the same eight examples.
Answer and interpretation
Both candidates survive training; their held-out predictions differ. The data establish this observed alternation pattern, not a unique grammar or every possible trigger.
Test the result
Keep suffix identity and surface sound separate. No vowel other than the explicitly specified i/e sets is classified by the candidates.
Change an assumption
Reveal the held-out environment
New exact observations are R1-e [pake] and R3-e [tege]. The original eight forms remain. Evaluate both deterministic candidates, then compare Hi with a rival that changes the stop only before this particular -i suffix, regardless of other environments.
- Both new forms contradict Hie, which predicted affricates. Under the stipulated no-exception model, reject that candidate.
- Hi predicts both stops and remains compatible with all ten forms. Its success does not test an i that belongs to some other suffix or inside a root.
- The suffix-specific rival also fits all ten. A new independently identified i-initial suffix would discriminate them: Hi predicts the alternation there; the rival does not. No such form is supplied.
Answer and interpretation
Hi survives and Hie fails, but Hi is not uniquely established. The remaining sound-conditioned versus suffix-conditioned distinction needs another carefully chosen environment.
Test the result
Do not invent the extra suffix’s observed form. State a prediction and a missing test, not a completed experiment or an actual-language rule.
Common mistakes
Spelling becomes sound
Use the specified IPA strings, not letter names.
Fit becomes proof
Two hypotheses can agree on every observed item and differ on an unobserved environment.
Survival becomes uniqueness
After one rival fails, other observationally equivalent analyses may remain.
Before moving on
- Choose a route for phonetic description, cross-language evidence or phonological inference.
- Check every paradigm cell and state discriminating predictions.
- Separate a surviving analysis from a uniquely identified grammar.
Sources and edition checks
Sources checked:
- Broeders/Gussenhoven: university open theory text
The university’s open theory text and pronunciation-practice companion serve different functions; respect each work’s credits and licence.
- Ladefoged/Johnson: selected seventh-edition publisher catalogue
The publisher catalogue identifies the selected 2015 seventh edition and print ISBN; companion access is a separate question.
- Odden: Cambridge second edition
Cambridge identifies Odden’s 2013 second edition; phonological analysis complements rather than replaces sound description.
These are proposed study activities, not official course assignments or a claim that every book has been read cover to cover.