Acoustic Learning, Inc.
Absolute Pitch research, ear training and more

Phase 19:  Solving the Puzzle

Preface to Phase 19:  Solving the Puzzle

I think I have an answer.

I think I know how adults can learn perfect pitch.

By learning music as a written language, an adult should be able to learn perfect pitch.

Of course, having an answer is not the same as having a solution.  I believe I understand what needs to be learned, and how it can best be learned—but I don't yet know how to put the elements together, which means I don't have an actual curriculum.  Nonetheless, I want to share what I know so that, if you're inclined, you can tackle it yourself.

It's not going to be "easy".  It shouldn't be difficult, because the approach will respect natural learning and development, but you won't be able to just pick it up and put it down again for a few minutes every day.  Learning any language—really learning it, and becoming fluent and literate—requires your full attention, maybe even to the point of immersion.

Missing from this explanation will be the most of the actual scientific references that support what I'm writing about.  Right now my goal is to explain what I know; although I have support for what I assert, for this round of writing it'll be too much of a slog to dig back in and re-locate every little scrap of research whose results have been residing in my brain all this time.  If you're skeptical about any claim that I make, please feel free to drop me a line and ask.

November 19 – The jigsaw is up

Why would anyone need perfect pitch?

If you only ever listen to other people making music, you certainly don't need perfect pitch. It doesn't matter what "pitches" they use. You can simply enjoy what you hear.

If you learn to reproduce other people's music, you still don't need perfect pitch. You figure out what notes to play on your instrument, you play them back, that's it. You may play with more or less skill, with more or less expressivity, but you don't need perfect pitch for that.

You don't even need perfect pitch to create your own music. Anyone can sing a random song of their own invention. You can think of a tune and noodle around on an instrument until you find its notes. You can make a melody, add a harmony, and not care which pitches are which.

Nobody needs perfect pitch to be an excellent musician.  That's why there's never been any imperative for musicians to learn it.  A musician may look at a page and hear, in their head, an accurate version of the notes printed there—just transposed. An instrumentalist can mentally map printed notes to their instrument without having to know any of the absolute pitches. And, if a composer knows what key they're composing in, they can put the correct notes on the page because their mental concept remains structurally the same whatever the actual pitches. You can perform and compose, proficiently, without perfect pitch.  This is why many people dismiss absolute pitch as nothing more than a novelty and a party trick: related to music, but not relevant to music.  But there is one task that only people with perfect pitch can do.

What you can't do without perfect pitch is read and write music.

Yes, I know that sounds preposterous. Of course, there are plenty of people who compose music without perfect pitch, as I've already said.  And, indeed, without perfect pitch, composers can conceive the shape of a melody. They can develop the structure of a harmony. But here's the wrinkle:  when they want to write down these ideas, they can't be sure that what they're writing is the actual sound of their creation.  The process of composing music is not the same as the specific cognitive actions of reading and writing. 

Reading and writing are the acts of matching mental sounds with printed glyphs. You're reading this, right now; that means these words are appearing in your head as sounds. Because you can hear these sounds in your imagination, you can speak these words out loud and know you're getting the sounds exactly right. You can also write whatever someone says out loud; you can transcribe anything, accurately and easily, because you know precisely which letters match the sounds you heard.  You don't need to hear a "reference letter" before you can read or write the correct sounds.  Reading and writing are the processes of transforming symbols into sounds, or vice versa, without any external aid.  Without absolute pitch, a musician can't do this.  Reading and writing require absolute pitch.

I don't claim that anyone needs perfect pitch.  What I am saying is that the reason to learn it is to become able to read and write music. Reading and writing music is the only purpose I know for which you must have perfect pitch.

And the purpose of reading and writing is to communicate ideas.

November 26 – Phoning it in

What is absolute pitch?

Defining a problem determines how we try to solve it. And, for more than a century, absolute pitch has been defined as "the ability to name notes". Consequently, for more than a century, people have trained themselves to name notes, and, consequently, for more than a century, no adult has learned perfect pitch.

The story of failure has been the same since 1899:

1. Someone designs a method to teach note naming.
2. A number of students follow the method.
3. A percentage of those students learn to name some notes.

The training method is irrelevant. The result is always the same. Whether it's association with colors, representative melodies, note "feelings", fixed scales, drug treatments, rote repetition, whatever, it doesn't matter: a few students learn to name some notes. Always. Every time. Then, typically, the designers point to the few successes as evidence that their method works, and they may invent reasons why the majority who failed were deficient (and therefore unable to learn). But the "successful" learners have not learned absolute pitch. All they can do is name notes in isolation. They can't name notes they hear in musical context. They can't apply note-naming to their musical practice. Every training method has always produced this same result, this same failure to learn absolute pitch, because every method has been designed to teach note naming. As long as the training goal continues to be naming notes, absolute pitch training will continue to fail.

Anyone can learn to name notes. But naming notes is not absolute pitch any more than holding handlebars is riding a a bicycle. Naming notes gets all the attention, but there's clearly more to it: Absolute listeners can perform many musical tasks more skillfully than those without (see Diana Deutsch's research), but they can be baffled by relative-pitch tasks which others find effortless (look up Ken Miyazaki). And everyone, regardless of musical ability, has at least some innate level of absolute pitch sense (just ask Dan Levitin). So we do know there is more to absolute pitch, but the reason we know this is not because absolute pitch has been defined. Rather, research has focused primarily on defining the effects, results, and consequences of having absolute pitch. There are discussions of its causes, which trace its origins (either to genetics or early musical experience) and question whether the unique neuroanatomy of absolute listeners is a cause or an effect, but these discussions don't contribute to definitions of absolute pitch. Where a definition of absolute pitch is offered, the definition appears as a catalog of the tasks we know absolute listeners do differently (behaviorally and neurologically); and, seated prominently at the head of that list, is note naming.

But a meaningful definition of "absolute pitch" can't be a catalog of musical tasks. Absolute pitch must be an overall cognitive framework for perceiving and producing music. Every single thing that absolute listeners can and can't do has to be a natural consequence of that single framework. Listeners with absolute pitch have a different perspective of, a different way of thinking about, a different way of processing music. What needs to be learned is the absolute listener's perspective, that way of thinking, that way of processing music. Learning to name notes cannot be an end goal.  Someone who learns to process music from an absolute-pitch perspective would have, as a natural expression of that framework, along with every other skill and drawback associated with absolute pitch, the ability to name notes.  We need to know what that framework is—because, if we want to learn absolute pitch, if we want to find a solution that works, we have to define this problem.

We can define an absolute-pitch framework through a linguistic perspective of music. No one would seriously disagree, these days, with the assertion that music is like a language. I argue that music is a language, because we use it to communicate musical ideas. Musical ideas are not literally comprehensible, as spoken-language ideas are—but they are, nonetheless, ideas. Musical ideas can be conceived, expressed, and understood. And there is a direct structural parallel between how musical and linguistic ideas are expressed. Language and music both have phrases. Where language has words, music has progressions. Words are made from syllables; progressions are formed with chords. And where the pieces of a syllable are phonemes, the elements of a chord are pitches. More succinctly:

In language, phonemes → syllables → words → phrases.
In music, pitches → chords → progressions → phrases.

This parallel implies a distinct musical function for absolute pitch. This function is as meaningful, relevant, and integral to musicianship as phonemes are to language. Namely:

If pitches are musical phonemes, then absolute pitch is musical phonemic awareness.

And that's what we need to learn.

December 7 – Structurally sound

What, then, is phonemic awareness?

This isn't hard to answer.  The very first search result provides a comprehensive definition.  (I've copied their words below just because they happened to be the first search result, but I could have drawn from anywhere.)  The definition of phonemic awareness is not just a list of tasks that people with the ability can do.  Rather, the definition describes the language-processing framework we call phonemic awareness.

Specifically, phonemic awareness is:

1.  The ability to hear and manipulate the sounds in spoken words and the understanding that spoken words and syllables are made up of sequences of speech sounds.

2.  Fundamental to mapping speech to print. If a child cannot hear that "man" and "moon" begin with the same sound or cannot blend the sounds r, u, n into the word "run", he or she may have great difficulty connecting sounds with their written symbols or blending sounds to make a word.

3.  Essential to learning to read in an alphabetic writing system, because letters represent sounds or phonemes. Without phonemic awareness, phonics makes little sense.

4.  A strong predictor of children who experience early reading success.

By substituting words in this definition, we could say that absolute pitch is:

1.  The ability to hear and manipulate the sounds in audible music and the understanding that musical progressions and chords are made up of sequences of pitch sounds.

2.  Fundamental to mapping music to print. If a child cannot hear that C-major and C-minor begin with the same pitch or cannot blend the pitches G, B, D into a G-major chord, he or she may have great difficulty connecting pitches with their written symbols or blending pitches to make a chord.

3.  Essential to learning to read in a musical notation system, because notes represent pitches. Without absolute pitch, pitches make little sense.

4.  A strong predictor of children who experience early reading success.

Absolute pitch is structural processing, just like phonemic awareness.  That's the essence of this definition (expressed in points one and two).  If absolute pitch were nothing more than the ability to hear individual pitches—especially as an automatic and unconscious process—then absolute listeners wouldn't be able to hear music.  They would hear no harmonies, no melodies, no musical sounds; they'd only perceive agglomerations of unrelated pitches.  But absolute listeners do hear music.

To understand what absolute listeners hear, think about how you yourself listen to language.  You aren't consciously thinking about all the letter sounds as they go flying past—but you're aware of them.  You know that what you're hearing is "made up of sequences of speech sounds."  If the speaker stopped and asked you to write down what you just heard, you'd be able to do it easily, and that's not because you were tracking every single one of the letters in their utterance.  You remember the meaning of the sentences, and the words that formed the sentences, and you know the spelling of those words, and you can therefore write them down.  This multi-level mental framework, this structural framework, is what makes sense of point two.  On the face of it, it seems silly to think that anyone couldn't "blend the pitches G, B, D into a G-major chord."  All you'd have to do is sing the pitches, one by one, and hey presto you've got the chord.  But stop a moment.  Think about phonemic blending.  There's a tremendous difference between someone who merely verbalizes the nonsense sounds "rrrr-uuuu-nnnn" and someone who, given those three phonemes, deliberately says the meaningful word "run".  Consider that mental difference for a moment.  Now apply it to music.  If you play G, B, and D for a musician with absolute pitch, they will recognize that you've created a G-major chord; and, if you ask them to sing it back, they won't sing three meaningless pitch sounds.  They will sing a G-major chord.  This mental process of musical production is directly parallel to language production.  When you write (or type) language, you produce individual letters, but you aren't thinking about the letters one by one.  You're thinking about the words and retrieving their letter sequences from memory.  Likewise, when someone with absolute pitch writes music, they're not thinking about individual pitches; they're thinking about the chords and retrieving their pitch sequences from memory.  Flipping this around, from writing to reading, is also parallel.  Reading language, you see individual letters, but you combine them into words; you see r, u, n and apprehend the word "run".  An absolute reader sees the individual pitches, but combines them into chords; they see G, B, D and apprehend the chord "G-major".  This is what it means to "hear and manipulate the sounds in audible music."  This is what it means to "map music to print."  Having absolute pitch means knowing the structural composition of musical sounds.

Points three and four of the definition are, I think, self-explanatory.  As long as you understand that reading means "looking at symbols and correctly hearing their sounds in your head", it should be obvious that absolute pitch is essential for reading music.  And just as children who read language become gradually able to read increasingly sophisticated literature, so too will children with absolute pitch become able to read increasingly sophisticated compositions.

I realize that I'm making important claims here without having absolute pitch myself—so I ran what I've just written past Bruce, a friend of mine who does have absolute pitch.  He affirmed,

My experience is consistent with most of what you've said here… when I hear a G major chord, I am perceiving the individual pitch sounds in an automatic way and can immediately identify the specific pitches (i.e., absolute pitch). Simultaneously, I am also hearing the gestalt of those pitches together and processing their sound as a chord à la relative pitch.

Bruce's description reminds me that the supposed dichotomy between absolute and relative pitch is, I think, a myth.  Any musician can identify a major chord, whereas only certain people know it's a G major chord; therefore, it seems reasonable to say that these certain people must therefore be using "absolute pitch" to recognize the G identity.  But it is a fallacy to conclude that this "absolute pitch" skill is distinct from, and that it is somehow being used instead of, "relative pitch".  Pitches and chords are interdependent.  Saying that someone's "relative pitch" suffers because of their "absolute pitch" perception is like saying that you become less able to recognize words if you can also hear their individual phonemes.  I don't see how pitches and chords could be separated from each other, in perception and in practice; each informs the other reciprocally.  Knowing the absolute pitch identity of a musical structure, and knowing the relative musical structure formed by a combination of pitches, are expressions of musical phonemic awareness.  "Relative" perception is anchored and defined by "absolute" perception.  The two are inextricable.

In sum:  absolute pitch, or musical phonemic awareness, means possessing a mental hierarchy of structural understanding, spanning pitch to composition and all the relationships in between.

December 29 – Listing to both sides

The definition of absolute pitch should be an overall cognitive framework.  It should not be a list of tasks that people with absolute pitch can perform.  A list of tasks, no matter how exhaustive, is a result of having absolute pitch, and doesn't explain anything.  A cognitive framework that makes it possible to perform the tasks is a cause, and the correct framework should explain everything.  If the cognitive framework of absolute pitch is that of processing pitches as though they were musical phonemes, and music as though it were its own language composed of those phonemes, then absolute pitch should, therefore, be (re)defined as phonemic awareness in music, or pitch awareness.

When I say that pitch is a "musical phoneme" I mean musical.  I am not asserting any overlap between language and music.  Specifically, this framework I'm proposing has nothing to do with the tonal-language hypothesis.  People who speak tonal languages do not process music in the same way that they process language (the brain-scan data show it).  It is possible, yes, that tonal languages may facilitate absolute pitch development, but absolute listeners' cognitive framework is unique to music.  Absolute pitch is parallel to phonemic awareness in its function, and in its behavioral consequences, but pitches are not phonemes, and language is not music.

If pitch awareness is parallel to phonemic awareness, then each musical ability attributable to absolute pitch should be parallel to some linguistic ability observable in literate adults.  Therefore, I should be able to take a list of absolute-pitch abilities and find hypothetically comparable phonemic abilities for each item on that list.  I say "hypothetically" because, scientifically, such comparisons are wild speculations.  As I said, the essential comparison I'm making (pitches to phonemes) should not be taken to suggest that they are the same thing or that our brains process them the same way.  We do not know whether the neurological activity is the same.  We don't know whether the physical inputs are sufficiently similar. There are no data validating these comparisons, so it is entirely possible that I'm making it all up in a fantasy of imagined connections that are ultimately not true.  This means the comparisons I'm about to draw are suggestive, not conclusive—but to me they've formed a pretty strong suggestion.  Strong enough, indeed, to think that teaching perfect pitch as though it were reading and writing a new language may be the right approach.  So I'm going to address a list of some of the most commonly recognized absolute-pitch abilities and, by drawing parallels to phonemic abilities, generate a more practical definition of absolute pitch.

Naming sounds is the most obvious parallel.  Absolute listeners can name pitches; literate adults can name phonemes.  This is, of course, more than just "naming" a given sound; it means categorizing that sound and retrieving the label associated with that category.  This is why I've always rejected claims that "semitone errors" are acceptable when evaluating someone's absolute pitch ability.  When you hear a clearly spoken, unambiguous phoneme, there's no chance you'll mistake it for a different one.  "Oh" is not "ooh" is not "uh", and you know it.  When a note is sounded clearly and in tune, and a listener names it incorrectly with a "semitone error", they are not demonstrating categorical perception of that pitch.  Categorical perception of phonemic elements is crucial.

Naming the key signature of a song is typically included on lists of absolute-pitch abilities—but it shouldn't be.  Any musician can find the key of a musical piece; what's more, we all do it the same way, whether or not we have absolute pitch: we infer the tonic from structural cues.  The difference, as Terhardt and his colleague determined in 1983, is just what you'd expect, which is that absolute listeners recognize the pitches of the structural cues while non-absolute listeners "feel" their way through them.  In fact, that's exactly what I did, just now.  I hummed to myself the first vocal line of the Stampeders' "Sweet City Woman" and, starting from the last note of that line, I sang two ascending semitones, and stopped, because that note felt like a tonic.  I looked up the sheet music online and, sure enough, the line ends on E and the key is F-sharp.  Was I actually singing an F-sharp?  No idea.  But I found the tonic.  Finding a key signature is a musical task, not an absolute-pitch task.  Anyone can do it.  The absolute-pitch skill is merely that of naming the tonic pitch once it's found.  In that regard, naming the key of a piece—naming a tonic after it has been found—is exactly the same as naming any old pitch.  Including "name a key signature" on the list of tasks that absolute musicians can do is, therefore, redundant and, I'd even say, misleading, as key-signature detection is not special to absolute pitch.

Next is a simple listening test:  the "nonmusical paradigm for identifying absolute pitch possessors" (Ross, Olson, Marks & Gore, 2004, Journal of the Acoustical Society of America, 116).  For this "paradigm", a test tone is played, followed by interfering tones.  Then, a new tone sounds, and the listener decides whether it is or isn't the same as the test tone.  This is ridiculously easy for absolute listeners (and difficult for everyone else).  It's obvious how easy this is if you think of the phonemic parallel: if you heard a phoneme, then a mishmash of nonsense syllables, and then a new phoneme, such as "uh… gorblyfligmooper… oh," and had to decide whether uh was different from oh, of course you'd answer correctly.  It would be silly to think you wouldn't.  This is because, to you, the "interfering" sounds are irrelevant.  When you hear the first phoneme, you don't try to hold an exact image of that sound in your short-term memory; rather, you immediately identify it, by applying a label to it, and it's that label you keep in mind.  When you hear the second phoneme, you label that second phoneme, too, and you answer "same" or "different" depending on whether the labels are the same.  You do not compare the actual sounds.  This is how absolute listeners succeed at the tone test.  We know this because when the tones are different, but are in the same pitch category (i.e., they have the same label), and the only way to succeed is by holding the actual auditory image in short-term memory, absolute musicians don't score any better than non-absolute musicians.  Ross et al. called this a "non-musical" test because test-takers don't have to verbally produce labels for the notes, but I think they're mistaken.  Their main conclusion is that this test should make it possible to "discover a non-musician who is able to perform accurately on this test."  But even in the unlikely event such a person were discovered, this person wouldn't be someone with absolute pitch.  It would be someone with a supernormal capacity to hold the image of a sound stimulus in their short-term auditory memory—and that's not absolute pitch.  Absolute listeners pass this test because they identify pitches using categorical labels, and that requires musical experience.  Unless you were a musician, there would be no reason to label pitch categories; and, without a label, there would be no way to represent the categories.  So you've got to be a musician to pass this test using absolute pitch, meaning that this "non-musical" test is, necessarily, musical, in exactly the way the researchers thought they were avoiding.  (This makes me wonder if illiterate adults would fail the phonemic version of this same test.)  Nonetheless, this test does affirm a core attribute of absolute pitch:  as with phonemic awareness, pitch awareness means having an abstract and symbolic mental representation of each individual sound—a "label"—that is separate from the sound itself.

Absolute listeners can sing a named pitch, on demand, without a reference.  You can say out loud any one of the phonemes you're looking at, right now, without a reference.  This is also a result of your mental association between an abstract symbol and a sound category stored in your memory.  It's an association you gained by learning to read.

Identifying all the pitches in a chord, or a "tonal mass", is like a musical spelling bee.  With phonemic awareness, we have a mental library of all the sounds we could be listening for.  That means we can "decode" any word, familiar or unfamiliar, into its consitutent letters.  If someone says ghlbtsk, we can capture the entire sound form in our short-term memory, as a "nonsense word", and mentally sift through it, piece by piece, because we know how to tell one piece from another.  That is, we recognize the unique identity of each particular bundle of sound, distinct from the rest of the mass, and can mentally segregate it.  Likewise, we can play a "nonsense chord" for an absolute listener, and it's just as easy for them to mentally sift through it, piece by piece, and "pull out" the discrete bundles they recognize as the individual pitches.  A non-absolute listener has difficulty detecting the pitches in a nonsense chord because they don't know what sounds they could be listening for, so they need to figure out how to distinguish the relative interactions from the individual sounds and, furthermore, recognize that the sounds they do manage to pick out are, in fact, pitches.  An absolute listener doesn't have this problem.  Absolute listeners can recognize what is and isn't a pitch because, through experience and learning, they have developed a mental library of auditory images.

Some absolute musicians are able to accurately reproduce a song after having heard it only once.  I don't know if this is true of absolute listeners generally; the only person I know of who is documented as being able to do this is Derek Paravicini, and he's a whiz-kid musical prodigy regardless of his absolute pitch.  We do know that absolute listeners do have better auditory memory (thanks to Deutsch and Dooley, 2013), but this phenomenon of instant song replication, if it is an absolute-pitch skill and not merely a heightened memory capacity, should be explicable as a function of pitch awareness.  And there is a direct phonemic parallel:  look once at this string of letters, then look away and see how many of them you can remember.

g p q t s p c k g b r l r d j f k x p w m r n d w h p f x z j k g e u r w

Now look once at this string of letters, then look away and see how many of them you can recall.

the quick brown fox jumped over the lazy dog

It's the same quantity of letters.  Obviously, though, you find it easier to remember them all.  This is because, of course, you're not remembering the letters.  You're remembering the words.  In fact, if you're as familiar with the sentence as most of us are, you don't even have to remember nine words; you just need to remember the one sentence, and decoding the entire thing into its individual letters doesn't tax your memory at all.  The larger your mental grouping (from words to phrases to sentence), the less information you have to hold in short-term memory.  Your phonemic awareness allows you to easily "remember" every one of the letters you heard by decoding each group into its smaller parts.  If absolute listeners are able to quickly reproduce songs, I suspect that this same mechanism, of grouping and decoding, is how they do it.  We don't know that for sure, because no one's tested it, but it would be simple enough to test, in the same way that I just tested your phonemic memory; we could play tone sequences for them that do or don't "make sense"—sequences which can or can't be meaningfully formed into larger groups—and see whether that makes a difference in how much they're able to recall.  If this mechanism is what's at work, then that's another notch for pitch awareness as the essence of absolute-pitch skill, as pitch awareness allows absolute listeners to remember more by reducing the amount of information they have to remember.

Having addressed a short list of absolute-pitch abilities, we should also reckon with its purported "inability."  Absolute pitch inability is essentially a myth, because having absolute pitch isn't inherently a musical handicap.  Rather, absolute pitch provides additional strategies for making otherwise "relative" judgments, and, in certain contexts, these alternative strategies can fail.  In a way, this is the entire issue: without absolute pitch, musicians wouldn't be able to develop or use these strategies, so that's how absolute pitch can be a hindrance, case closed.  But, in re-reading Ken Miyazaki's experiment on transposed melodies, it occurred to me that there's an interesting comparison to phonemic awareness that I don't want to overlook.  I think this comparison can give some insight as to why and how absolute listeners would persistently apply an absolute strategy.

An absolute strategy for transposing melodies strikes me as nearly identical to most people's strategy for changing their speaking accent.  Musical transposition is a struggle for absolute musicians when they process it sound by sound, laboriously changing them one at a time, instead of making one adjustment to key signature and working from that new basis.  As I have explained in my TEDx talk, accent modification is a struggle for almost everyone because they process it sound by sound, laboriously changing them one at a time, instead of making one adjustment to articulatory settings and working from that new basis.  What makes it even worse is that, despite knowing what sounds they want to produce, their brain keeps overriding their intention and producing the "correct" original sounds instead.  This is one case where we can see phonemic awareness as an "inability".  In trying to talk with a different accent, speakers apply an "absolute" strategy (of changing each sound), even though a "relative" strategy (of moving to a new articulation) would be more effective.  And why wouldn't they?  Why would anyone even conceive of a "relative" strategy to change their accent?  The sound-by-sound approach is a logical, obvious conclusion, thanks to your phonemic awareness:  you know that words are made of phonemes, and you know that phonemes change when you speak with an accent—so it naturally follows that, to speak with an accent, you must change the phonemes.  And yet you can't do this successfully unless you overcome your phonemic awareness.  To illustrate: as an American, you know that "laugh" is composed of l, æ, and f.  When you think about saying "laugh", you automatically assemble those specific phonemes.  To say it as a British speaker would, you have to aggressively contradict what you know.  Having thought of saying "laugh," you have to stop yourself from saying "laugh," mentally replace the phoneme æ with ɒ, and then say the wrong sounds.  What a struggle!  Without phonemic awareness, you don't have this problem.  Some years ago, I was walking through a Party City store when I overheard a young girl, no more than four or five years old, exclaim "It's-a me, Mario!"  But she didn't say "It's-a me, Mario."  She said "eets-a me, moddy-oh," which is how the voice actor performs it.  She was at an age where her reading skills were still developing.  She knew the meaning of what she was saying; she just didn't yet know that "it's" begins with ih, not ee, and that "Mario" has an r, not a d.  If her phonemic awareness had been stronger, she would have said the "correct" words, using the correct phonemes.  Instead, she repeated the words as she had actually heard them.  Now, I don't know whether transposing a melody is physically or cognitively the same as speaking with a different accent.  It probably isn't.  I just think that the strategy most people use for speaking with an accent—a strategy which logically arises as a natural consequence of phonemic awareness, but to which phonemic awareness is itself a hindrance—can help us understand why a musician would invent and use an absolute strategy even when it doesn't work as well as a relative strategy would.  If you think about the accent-speaking strategy analogously with music transposition, imagining someone with pitch awareness trying to override their mind's insistence on producing the "correct" musical sounds, it's clear why and how pitch awareness could operate as an "inability".

Finally, a word about the supposed irrelevance of absolute pitch.  No one would ever dream of saying that phonemic awareness is unimportant or irrelevant to language use.  How could anyone say such a thing about pitch awareness in music?  Well, if the only reason you need absolute pitch is for reading and writing, it's easy to answer anyone's assertion that absolute pitch is unnecessary:  just convert it to a statement about language literacy.  For example, here is a quote from Miyazaki's 1992 paper:

Absolute pitch possessors can [identify]… a tone presented in isolation.  However, an isolated tone without musical context has no relevance to music.

Compare that to an equivalent statement about phonemes and language:

Phonemic awareness possessors can identify a letter sound presented in isolation.  However, an isolated letter without linguistic context has no relevance to speech.

This second statement is entirely true (the isolated letter g certainly doesn't mean anything), but no one is using it as an argument that we shouldn't have phonemic awareness.  Any statement claiming that perfect pitch is pointless can be converted in the same way; just imagine that the person is talking about language use and doesn't care about reading and writing.  Whatever the statement may be, changing it to a language context should make clear that it isn't really an argument against having pitch awareness, especially when you remember that reading means hearing sounds in your head and not mechanically transliterating from page to keyboard (or fret, or valve, or whatever your instrument has).  You don't need to read or write to be able to use language proficiently.  If reading and writing is not important, you don't need phonemic awareness.  For the same reasons, you don't need absolute pitch to be a musician.  But that doesn't make it useless.

A new definition of absolute pitch can turn our perspective around, turning focus away from what absolute listeners can do and toward what absolute pitch actually is.  It can stop us wondering how to mimic the results of having absolute pitch and start us discovering how to replicate its causes.  I think a starting point for a helpful new description of absolute pitch can be:

•  Categorical perception of pitches
•  An abstract and symbolic representation of each pitch, separate from the sound
•  A mental library of auditory pitch images
•  The understanding that tonal structures can be decoded into their component pitches

And, together, these help define absolute pitch as pitch awareness.

January 6 – A shade too much

There is one essential difference between absolute and non-absolute pitch perception.  We've known this for years.  Non-absolute listeners perceive pitch in one dimension ("height") whereas absolute listeners perceive two ("chroma" and "height").  You might think, therefore, that acquiring perfect pitch would just be a matter of learning to hear that second dimension of chroma; and, for decades, people have been trying.  The success of the Ear Training Companion, unfortunately, revealed that hearing chroma is not enough.  We still need to learn how to perceive it absolutely—and our ability to learn that is crippled by our existing concept of pitch perception.

Non-absolute listeners perceive pitch as a single dimension.  We call that one dimension "height", and the range of human hearing is from a "low" of 20 Hz to a "high" of 20,000 Hz.  Visually, we can represent the single dimension like this:

20 Hz Pitch height 20,000 Hz

We can't learn to name pitches because this one-dimensional range is a continuum:  "a continuous sequence in which adjacent elements are not perceptibly different from each other."  Our perceptual system requires differentiating features to identify objects (cf. Gibson's Principles of Perceptual Learning, 1969) and, within a continuum, by definition, there aren't any.  Different "heights" do not differentiate pitches, because these are magnitudes of the same feature.  It's like trying to learn an exact length.  You might be able to remember eighty-three centimeters, with repeated practice, but how can you know it's not eighty-two centimeters without the help of an "external reference" (e.g., a ruler)?  You can't.  Likewise, with repeated practice, you might be able to remember the sound of an F-sharp, but you can't ever be sure you're remembering it correctly because there's no differentiating feature to latch on to that confirms it.  Because a continuum is a single dimension, with an infinite number of individual elements, all defined by a single common trait, there is nothing distinctive about any one element that will allow you to identify it.

You could try to identify points in a continuum by making associations with phenomena outside the continuum.  You could, for example, make a subjective judgment that a certain musical pitch "feels mellow" or that another pitch "sounds intense".  You might draw a real-world connection, such as trying to identify a certain level of brightness because it is identical to the output of a standard light bulb.  Or you could choose a personal association; if you were a fan of old vinyl, and knew very well the feeling of an LP between your hands, you might use that sensory memory to decide how long twelve inches is.  Associations like these always seem valid and helpful.  They can help you make accurate identifications, and, after some practice, you could even have better-than-chance success, which would give you the impression that you have started to learn to identify the target values.  But that impression is an illusion.  When you use an external association like this, you're not identifying a stimulus.  You're recalling a wholly separate sense memory and comparing it to the stimulus, trying to figure out whether or not there is a match.  Sometimes there is, and sometimes there isn't.  But, either way, you're not learning anything about the stimulus.  Imagine trying to learn the color red by making an external association.  Maybe you decide that it "feels hot".  But orange and yellow can also feel hot, to different degrees, and a blue flame is hotter than a red one, and your ears and nose can turn red with cold.  Using a "hot feeling" association would help you recognize red in many contexts, but it wouldn't explain why you sometimes mis-identify the other colors as red, or why you sometimes don't recognize red because it doesn't feel hot.  In short, it doesn't and can't help you understand why red is distinct from orange, or yellow, or blue.  Any such association would have similar problems.  You don't learn to identify a color because of a distinct attribute you impose on it; you learn because of the distinct attribute it imbues.  It is the same with pitch sounds.  You can't learn to identify pitches by making external associations because the distinctive feature you're identifying is from your memory of something else.  It is not drawn from the pitch.  And it couldn't be.  Within a one-dimensional continuum, pitches do not have distinctive features.  Pitches merely have different positions on the "height" line.

"Height" is the defining characteristic of a pitch—and it is irrelevant.  It has no meaning until we impose onto it some structure.  We may call this structure a contour, and, of course, we can place a contour anywhere on the "height" continuum.  Once we have placed our contour, every pitch in the continuum assumes its identity from its function relative to the contour.  The contour, and the relative functions of its pitches, is all that matters to the identity of musical sound.  It's like looking at these flowers:


If you're asked are these the same flower, the answer is of course it is.  It just has different lighting levels.  We could, if we wanted to, hold up each version to a reference and determine exactly what level, but that's just a technical exercise.  Only the shape is definitive.  The problem in music, of course, is that, if the contours are identical, we can't perceive any pitch-level difference.  Contour is all we perceive, because we only identify the pitches through the relative functions they fulfill in forming the contour.

Our indistinct one-dimensional perception of pitch level is tellingly similar to the unusual abnormal case of the fellow who had never learned to identify colors.  This person perceived colors "entirely as intensities and not as qualities."  To him, colors were not different from each other.  He could only describe each color as having a level of "brightness" (although what he experienced as brightness appeared to be what we would call saturation).  In the same way, we don't recognize that pitches have different qualities, but only different intensities, so we tell the difference between them from the "distance" between their intensities—and "distance" is a measurement, not a pitch quality.  A seven-semitone "distance" would be exactly the same regardless of where it occurred on the pitch continuum.  So when we play tones in sequence, we "move" across the "distances" between them, which produce different harmonic tensions in the key context (and with each other), and create a melodic "contour"… and we hear exactly these same features regardless of pitch "height."

We do not percieve differences between pitches.  Without a relative context, all points on the continuum are functionally identical.  All pitches are perceptually the same.

Is it any wonder that every attempt to learn note-naming has failed?

We have to stop trying to name notes.  As long as we perceive pitch as a single, one-dimensional continuum, it can't be done.  To learn absolute pitch, and successfully become able to identify pitches, we first need to perceive pitch as an absolute listener does.