Lexical Tone and Pitch Accent
Chapter Objectives
By the end of this chapter, you will be able to:
- Distinguish lexical tone from intonation.
- Distinguish level tones from contour tones.
- Interpret one declared tone-notation system used consistently within an exercise.
- Recognize tone sandhi as a system-specific process rather than a universal rule.
- Explain why pitch realization varies by speaker and context.
- Analyze a controlled tonal dataset.
An Observable Problem
Chapter 20 established that stressed and unstressed syllables in English are frequently distinguished by a combination of duration, intensity, and pitch, among other cues. Pitch, however, can do far more work in a language than simply marking which syllable is prominent. Consider a hypothetical case: a single syllable, held constant in every other respect, produced with a level high pitch means one word, while the same syllable produced with a pitch that starts low and rises means a completely different word. No stress difference is involved. No segmental difference is involved. Only the pitch shape changes, and the meaning changes with it.
This is not a hypothetical curiosity. In a large proportion of the world's languages, pitch operates this way as a normal, everyday part of distinguishing one word from another, in much the same way that English speakers rely on distinguishing p from b. A monolingual English speaker encountering such a system for the first time often assumes that the pitch differences are a kind of emotional coloring, similar to the rise a question receives in English. That assumption is a category error this chapter is designed to correct.
This chapter asks three questions:
- How does a pitch contrast that changes word meaning, called lexical tone, differ from the phrase-level pitch patterns introduced briefly in Chapter 20 and developed fully in Chapter 22?
- What kinds of pitch patterns count as tones, and how are they represented in a declared, consistent notation system?
- Why can the same tone change its realization depending on neighboring tones or the speaker producing it, and what does that mean for how confidently a learner can generalize from a single example?
Lexical Tone Versus Intonation
Purpose: define lexical tone precisely and separate it from the phrase-level pitch patterns previewed in Chapter 20 and developed in Chapter 22, so that the two systems are never treated as the same phenomenon.
Lexical tone is the use of pitch, at the level of the syllable or word, to distinguish one word or morpheme from another. A language with lexical tone assigns a pitch pattern to a syllable as part of that syllable's phonological identity, the same way it assigns a particular vowel or consonant. Changing the tone changes the word, independent of any sentence-level pitch pattern layered on top of it.
Intonation, previewed here and developed fully in Chapter 22, is phrase-level or sentence-level pitch patterning that conveys information such as sentence type, focus, or discourse structure, and that applies across a stretch of speech rather than distinguishing individual lexical items. A language can have intonation without lexical tone, lexical tone without notable intonation contrasts, or, as is true of many tone languages, both systems operating at once, layered on top of each other.
A common misconception is that a tone language has no intonation, because pitch is already "used up" distinguishing words. This is inaccurate. Tone languages generally have their own intonation systems as well; the lexical tone pattern of a word and the intonation contour of the phrase it occurs in are layered together, and analyzing tone languages requires separating the two contributions rather than assuming one displaces the other. Untangling that layering in detail is a task for Chapter 22 and beyond; this chapter focuses on the word-level system alone.
A useful diagnostic, extending the diagnostic Chapter 20 used to separate lexical stress from phrase-level prominence, is to ask whether the pitch pattern is a fixed property of the word, unlikely to change across ordinary sentence contexts, or a pattern that depends on the sentence type, the speaker's discourse goals, or the position of the word in the utterance. A fixed, word-identifying pitch pattern points toward lexical tone; a pattern that varies with sentence function points toward intonation.
Not every language uses pitch to distinguish word meaning. Languages without lexical tone, including most varieties of English, rely on stress, as developed in Chapter 20, and on segmental contrasts to distinguish words, reserving pitch primarily for intonation and other paralinguistic functions. Do not assume that the pitch movements audible in any language are evidence of lexical tone; confirm this analytically, using minimal or near-minimal tonal pairs, before making the claim.
Level and Contour Tones
Purpose: distinguish the two broad shapes a lexical tone can take and connect them to a controlled, illustrative tone inventory.
Tones are commonly divided into two broad categories based on their pitch shape over the syllable:
- Level tones maintain a relatively steady pitch height across the syllable, such as a tone realized consistently high, consistently mid, or consistently low relative to the speaker's pitch range.
- Contour tones involve a pitch movement within the syllable, such as a tone that rises, falls, rises then falls, or falls then rises.
A single language may use only level tones, only contour tones, or, commonly, a combination of both, organized into a tone inventory specific to that language. The number of tones a language distinguishes, and which pitch shapes count as separate tones, are empirical facts that must be established for each language rather than assumed from a typological stereotype.
Illustrative pitch shapes and their approximate IPA tone letters and Chao-numeral equivalents, not a claim about any single language's full tone inventory.
Standard Mandarin is commonly described as distinguishing four lexical tones on most full syllables: a high level tone, a rising tone, a falling-rising tone, and a falling tone, alongside a phonetically reduced neutral tone in certain contexts. The classic minimal set built on the syllable ma is frequently cited to illustrate that changing only the tone changes the word; consult a scholarly source or dictionary for the specific glosses and characters before using this set in narrow instructional material, since exact gloss sets vary by source.
Notating Tone
Purpose: introduce the course's primary tone-notation convention, connect it to an alternative numeral system used for comparison, and prevent silent mixing of systems.
This textbook uses the official IPA tone letters, shown attached to a vertical reference line, as the primary convention for representing tone in transcription. The tone letters represent pitch height and direction iconically: a mark that is high on the line represents high pitch, a mark that is low on the line represents low pitch, and the slope of the mark represents a rising or falling movement.
| Symbol | Name | Meaning |
|---|---|---|
| ˥ | Extra-high tone letter | A level tone at the top of the speaker's pitch range for the relevant stretch of speech. |
| ˦ | High tone letter | A level tone above mid but below extra-high. |
| ˧ | Mid tone letter | A level tone near the center of the speaker's pitch range. |
| ˨ | Low tone letter | A level tone below mid but above extra-low. |
| ˩ | Extra-low tone letter | A level tone at the bottom of the speaker's pitch range. |
| ˧˥ etc. | Combined tone letters | Two or more level marks combined to represent a contour; the sequence traces the pitch movement across the syllable. |
| ꜒ / ꜓ | Downstep and upstep | Marks a step down or up in pitch register relative to a preceding tone, distinct from the pitch shape of a single syllable; introduced here for recognition and used productively only where a specific language's analysis requires it. |
Chao numerals, sometimes called tone letters in numeral form, provide an alternative convention. In this system, pitch height is represented on a scale from 1, the lowest, to 5, the highest, and a tone is written as a sequence of one or more digits tracing its pitch shape, such as 55 for a level high tone or 214 for a tone that falls and then rises. Chao numerals are common in the descriptive literature on tone languages, particularly for Sino-Tibetan languages, and this textbook introduces them only for comparison with the IPA tone letters, not as a replacement for them.
Do not mix IPA tone letters and Chao numerals within the same exercise or transcription. State explicitly which system is in use, and convert consistently if a task requires moving between the two.
Error: Representing tone with an informal device such as arrows, emoji, or ad hoc punctuation instead of a declared notation system.
Correction: Use the IPA tone letters or a clearly declared numeral system consistently. Informal devices cannot be compared across examples or languages and do not meet the transcription standards developed since Chapter 24 will formalize; establish disciplined habits with tone notation now.
Pitch Accent
Purpose: introduce pitch accent as a related but distinct word-level pitch system, positioned between the stress systems of Chapter 20 and the fuller tone systems described above.
Some languages are described as having a pitch accent system, a term used inconsistently across the descriptive and theoretical literature but generally referring to a system in which pitch marks a designated, often limited, prominent point in the word, rather than assigning an independent tone to every syllable the way a fully tonal language does. A pitch-accent system typically restricts the possible pitch patterns of a word far more than a tone system does, and the location of the accented syllable, rather than a full inventory of contrastive tone shapes, is usually the primary contrastive information.
Because the term is applied differently by different scholarly traditions, and because some languages described as pitch-accent languages have analyses that overlap substantially with either stress systems or tone systems, treat any specific pitch-accent claim as requiring the source's own definition of the term rather than assuming a single universal pitch-accent template.
Standard Japanese is commonly analyzed as having a pitch-accent system in which a word may carry an accent on a particular mora, marked by a fall in pitch immediately after the accented mora, or may carry no accent at all. This contrasts with Standard Mandarin, where every full syllable carries one of a small set of independently contrastive tones. Comparing the two systems illustrates that "pitch matters to word identity" is not, by itself, a specific enough description of a language's system; the details of how many contrasts exist and where they can occur must be established separately for each language.
Stress systems, pitch-accent systems, and fully tonal systems are sometimes presented as three sharply separate boxes, but many linguists instead treat them as points along a continuum of how much a language's phonology relies on pitch to distinguish words and how freely that pitch can vary syllable by syllable. Locating a specific language on this continuum is a claim requiring dedicated scholarly evidence, not a judgment a learner should make from a few isolated examples.
Tone Sandhi
Purpose: introduce tone sandhi as a system-specific contextual process, connecting it forward to the connected-speech processes of Chapter 23 without teaching it as a universal property of all tone languages.
Tone sandhi refers to a predictable change in a tone's realization that depends on the tones of neighboring syllables. A tone that is produced one way in isolation may be produced differently when it occurs next to a specific other tone, in a pattern that is regular and rule-governed for the language in question, much as the connected-speech processes introduced in Chapter 23 regularly alter segmental realizations in context.
Tone sandhi is a system-specific fact. Its existence, its triggering environment, and its precise effect must be established separately for each tone language; a sandhi pattern documented for one language, or even one variety of a language, cannot be assumed to hold for another. Treat any sandhi rule presented in this textbook as an illustration of the general concept, not as evidence that every tone language behaves the same way.
Standard Mandarin is frequently cited for a sandhi pattern in which a falling-rising tone, when immediately followed by another falling-rising tone, is commonly realized instead as a rising tone. This is a well-documented, language-specific rule rather than a general property of tone languages; other tone languages have entirely different sandhi patterns, and some documented tone languages show little or no sandhi at all. Confirm the precise environment and exceptions for any specific sandhi claim with a reliable source before teaching it as a firm rule.
Speaker and Context Variation
Purpose: prevent overconfident claims about pitch by connecting tone realization to speaker-specific and contextual factors, extending the variation policy applied throughout the course.
The physical pitch of a given tone, measured in fundamental frequency, differs substantially from speaker to speaker due to factors such as vocal-fold length and typical pitch range; a tone described as "high" for one speaker may fall in a different absolute frequency range than the same tone produced by another speaker with a different typical pitch. For this reason, tone is analyzed relative to a speaker's own pitch range rather than by a fixed frequency value shared across all speakers.
Tone realization can also shift with speaking rate, emotional state, sentence position, and the intonation pattern layered over a phrase, previewed earlier in this chapter and developed in Chapter 22. A description of a language's tone inventory captures the contrastive categories a speaker uses; it does not predict the exact pitch trace of a single spoken token in every context.
Do not treat a tone category as equivalent to a specific number of hertz. A high tone is high relative to that speaker's own range and to the other tones in that speaker's system, not relative to a universal pitch value. When acoustic evidence for tone is introduced in Chapter 28, fundamental-frequency measurements will be interpreted relative to a specific speaker and recording, following the same evidentiary caution applied throughout this book.
Across Languages
Yorùbá is commonly described as distinguishing three level lexical tones, high, mid, and low, marked in its practical orthography with diacritics rather than IPA tone letters. This system illustrates that a tone language need not rely primarily on contour tones; a purely level-tone inventory can still carry substantial lexical contrast. Consult a scholarly source before citing specific Yorùbá minimal sets in narrow instructional material.
Standard Vietnamese is commonly described as distinguishing six lexical tones, several of them contour tones that also involve differences in phonation such as creaky or breathy voice, previewed in Chapter 6 and Chapter 18. This shows that a language's tone system can interact with phonation type as well as pitch shape, reinforcing the point that a full tone inventory is an empirical package of properties specific to the language, not pitch shape alone.
Common Mistakes
Error: Assuming that any perceptible pitch movement in speech is evidence of lexical tone.
Correction: Pitch movement is present in the speech of every language, tonal or not, because intonation is a near-universal system. Lexical tone requires evidence that the pitch pattern is a fixed, contrastive property of the specific word, demonstrated with minimal or near-minimal tonal pairs, not simply the presence of audible pitch change.
Error: Treating "pitch accent" as a simple midpoint meaning "a little bit tonal" that applies identically to every language given that label.
Correction: Pitch accent is used inconsistently across scholarly traditions and covers systems with meaningfully different properties. Rely on the specific source's definition and description for any given language rather than assuming one universal pitch-accent template.
Error: Assuming a sandhi rule documented for one tone language, or one variety of a language, applies to every tone language.
Correction: Tone sandhi is a system-specific, rule-governed process that must be established separately for each language and variety. Some tone languages show extensive sandhi; others show little or none.
Chapter Review
Part A — Tone Versus Intonation
1. Define lexical tone and explain the diagnostic this chapter uses to distinguish it from intonation.
2. Explain why a language can have both lexical tone and intonation at the same time, rather than one displacing the other.
Part B — Tone Shapes and Notation
1. Distinguish level tones from contour tones and give one example of each using IPA tone letters.
2. Explain why IPA tone letters and Chao numerals should not be mixed within the same transcription or exercise.
Part C — Pitch Accent, Sandhi, and Variation
1. Explain how a pitch-accent system, as illustrated by Standard Japanese, differs from a fully tonal system, as illustrated by Standard Mandarin.
2. Explain why tone sandhi must be treated as a system-specific process rather than a universal property of tone languages.
3. Explain why a tone category is analyzed relative to a speaker's own pitch range rather than as a fixed frequency value.
Cumulative Review
1. From Chapter 20: Compare the diagnostic used to separate lexical stress from phrase-level prominence with the diagnostic used in this chapter to separate lexical tone from intonation. What do the two diagnostics have in common?
2. From Chapter 18: Explain how the Vietnamese example in this chapter connects lexical tone to the phonation concepts, such as creaky and breathy voice, introduced in Chapter 6 and expanded for vowels in Chapter 18.
3. Preview for Chapter 22: This chapter treated intonation only as a contrast case for lexical tone. Predict what additional information, beyond a simple rise or fall, a full account of phrase-level intonation might need to capture, a question Chapter 22 addresses directly.
Self-Check
Before moving to Chapter 22, confirm that you can do the following without looking at your notes:
- Distinguish lexical tone from intonation using a clear diagnostic.
- Identify a tone shape as level or contour from an IPA tone letter.
- Convert a simple tone between IPA tone letters and Chao numerals without mixing the two systems.
- Explain the general difference between a pitch-accent system and a fully tonal system.
- Explain why tone sandhi is analyzed separately for each language.
- Explain why tone categories are relative to a speaker's pitch range rather than fixed frequencies.
If any of these tasks are difficult, review the relevant section before reading Chapter 22. Chapter 22 shifts from word-level pitch to phrase-level pitch, building directly on the intonation preview introduced in this chapter.
Answer Key (click to expand)
Analysis Exercises — Exercise A
1. ˥: level tone, held at extra-high pitch across the syllable. 2. ˧˥: contour tone, rising from mid to high. 3. ˥˩: contour tone, falling from extra-high to extra-low. 4. ˨˩˦: contour tone, falling from low to extra-low and then rising to high.
Analysis Exercises — Exercise B
1. Intonation. The pitch rise appears only in the question context and disappears when the same word occurs in a statement, showing that the pattern depends on sentence function rather than being a fixed property of the word. 2. Lexical tone. The falling pitch shape is stable across sentence positions and sentence types for this specific word, matching the diagnostic for a fixed, word-identifying pitch pattern.
Analysis Exercises — Exercise C
1. ˧ converts to the Chao numeral 33, a level tone at mid height. 2. The Chao numeral 51 converts to the IPA tone letter sequence ˥˩, a tone falling from extra-high to extra-low.
Chapter Review — Part A
1. Lexical tone is the use of pitch at the syllable or word level to distinguish one word or morpheme from another; it is diagnosed by asking whether the pitch pattern is a fixed property of the word across contexts, pointing to tone, or a pattern that depends on sentence type or discourse function, pointing to intonation. 2. Tone and intonation operate on different levels, the word and the phrase respectively, so a language can layer a phrase-level intonation contour on top of the fixed lexical tones of its individual words rather than one system replacing the other.
Chapter Review — Part B
1. Level tones hold a steady pitch height across the syllable, such as ˥ for a high level tone; contour tones involve pitch movement within the syllable, such as ˧˥ for a rising tone. 2. Mixing IPA tone letters and Chao numerals within the same transcription makes the notation ambiguous and inconsistent, since a reader cannot tell which convention a given mark or digit belongs to; the chapter requires declaring and maintaining one system per exercise.
Chapter Review — Part C
1. A pitch-accent system, as in Standard Japanese, typically marks a single designated prominent point in the word with a pitch event, such as a fall after an accented mora, and restricts the possible pitch patterns of a word accordingly; a fully tonal system, as in Standard Mandarin, assigns an independently contrastive tone to each syllable from a larger inventory of pitch shapes. 2. Tone sandhi's existence, triggering environment, and effect are established separately for each language and variety; some tone languages show extensive sandhi, such as Mandarin's third-tone sandhi, while others show little or none, so no sandhi rule can be assumed to transfer automatically between languages. 3. Absolute pitch varies by speaker due to factors such as vocal-fold length and typical range, so a tone category is defined relative to a given speaker's own pitch range and the other tones in that speaker's system, not by a fixed frequency shared across all speakers.
Cumulative Review
1. Both diagnostics ask whether a pitch or prominence pattern remains fixed as a property of the word across different contexts, pointing to a lexical-level system such as stress or tone, or instead shifts depending on phrase-level factors such as focus or sentence type, pointing to a phrase-level system such as phrase-level prominence or intonation. 2. The Vietnamese tone system shows that some tones involve not just pitch shape but also a specific phonation type, such as creaky or breathy voice, connecting this chapter's tone inventory to the phonation concepts introduced for consonants in Chapter 6 and expanded for vowels in Chapter 18. 3. A full account of phrase-level intonation will likely need to describe how pitch is organized across an entire phrase or sentence, including boundaries between phrases, points of prominence within a phrase, and how these relate to discourse meaning such as questions, statements, or contrast, going well beyond a single rise-or-fall label.
Looking Ahead
What You Can Now Do
You can now distinguish lexical tone from intonation, identify level and contour tones, interpret IPA tone letters and convert them to and from Chao numerals without mixing systems, distinguish pitch-accent systems from fully tonal systems, recognize tone sandhi as a language-specific process, and explain why tone realization depends on the individual speaker's pitch range and context. This foundation in word-level pitch prepares you for the phrase-level pitch systems that occupy the remainder of Part IV.