When Words Combine
Japanese loves compound words - combining two or more words into one. When this happens, the pitch accent often changes according to predictable rules.
For example, 日本 (にほん, Japan) + 語 (ご, language) = 日本語 (にほんご, Japanese language). The pitch pattern isn't just the two words stuck together - it forms a new pattern!
McCawley's Rules
Linguist James McCawley identified patterns for predicting compound accent when a word isn't in the dictionary. The key factor is the mora count of N2:
Important: These are predictive guidelines, not absolute rules. Dictionary entries (like 日本語⁰, 英語⁰, 花見³) override any prediction. Mierutone uses dictionary data when available, and only applies McCawley rules for unknown compounds.
N2 has 1-2 moras → Predict accent on last mora of N1
For short second elements, predict the accent falls on the last mora of N1.
N2 has 3-4 moras → Predict accent on first mora of N2
Medium-length second elements shift the predicted accent to where N2 begins.
N2 has 5+ moras → Follow N2's original accent
Long second elements (5+ moras) tend to keep their own accent pattern, offset by the length of N1. This applies to longer technical terms and loanwords.
Example: コミュニティーセンター⁶ - N2 preserves its accent shifted by N1.
Dictionary vs Rules
McCawley's rules are predictive guidelines, not absolute laws. Many common compounds have dictionary entries with their actual accent patterns, which may differ from the predicted pattern.
How Mierutone handles this:
- • Dictionary match → Uses the recorded accent (high confidence)
- • No dictionary entry → Applies McCawley rules (lower confidence)
- • Shows component breakdown so you can see the analysis
Compound Structure Breakdown
The analyzer shows you how compounds are split and analyzed:
Common Compound Suffixes
Some suffixes appear frequently and follow consistent patterns:
language (1 mora)
person (2 moras)
-tic/-like (2 moras)
company (3 moras)
university (4 moras)
problem (4 moras)
Practice Tips
- 1.Start by learning common suffixes (-語, -人, -的) and their typical behavior.
- 2.When you encounter a new compound, try to identify N1 and N2 first.
- 3.Count the moras in N2 to predict where the accent might fall.
- 4.Remember: dictionary entries override predictions. When in doubt, check!
- 5.Use the analyzer to see both the compound pattern and component breakdown.
Exceptions & Variations
Not all compounds follow McCawley's rules perfectly. Some factors that affect accent:
- • Word origin (goshu): Native Japanese vs Sino-Japanese vs foreign loanwords
- • Frequency: Very common compounds may have crystallized patterns
- • Semantic bonding: How tightly the elements are fused
- • Regional variation: Dialects may differ from standard Tokyo accent
When the analyzer shows "compound_rule" as the source with low confidence, it's applying McCawley rules as a prediction.
Sources
- McCawley, James D. (1968). The Phonological Component of a Grammar of Japanese. Mouton.
- Kubozono, Haruo (2011). "Japanese pitch accent." The Blackwell Companion to Phonology.
- NHK日本語発音アクセント新辞典 - Standard reference for Tokyo dialect pitch accent
- Pitch data from Kanjium dictionary project.