Speech Segmentation and Lexicon Building
How do children find words in a stream of English?
In spoken English, there are not always clear spaces between words. Children segment continuous speech using cues such as sound patterns, stress, and syllable structure, then store the word forms they discover in the lexicon.
What Is Speech Segmentation?
When we look at written English, spaces separate one word from another. In actual speech, however, there are no such spaces. In fact, sounds link across word boundaries, change shape, and sometimes even disappear. Even so, English speakers can continuously hear words emerging from the flow of speech.
Once speech perception has developed enough to let children hear the sound patterns of a language in a language-specific way, the next skill they need is the ability to find words and phrases within a continuous stream of sound. This process is called segmentation.
Finding repeated chunks in the flow of sound
Babies do not begin life already knowing the meanings of words. Even so, when they hear large amounts of speech, they start to notice that similar sound chunks appear again and again.
For example, even before a child knows what dog means, repeated exposure to sentences such as Look at the dogs. Dogs are running. There is a dog. helps the child notice the recurring sound chunk “dog” within them.
At this stage, the first step is to detect recurring sound chunks in continuous speech and treat them as word-like units. That is speech segmentation.
Real speech has no spaces, but when the same sound chunk appears repeatedly, children begin to recognize it as a single word-like unit. Even without knowing the meaning, they can mark it as a candidate word form.
Predicting where a word begins from prominent sounds
Children who do not yet have enough meaning or vocabulary use multiple cues to figure out where one word begins and ends in the stream of speech. One of those cues is stress.
English words contain strong and weak syllables, and many everyday words follow a trochaic pattern, beginning with a strong syllable followed by a weak one. As a result, children who grow up hearing English begin to use strongly heard syllables as cues and predict, “Maybe a new word starts here.”
This tendency also appears in young children’s speech. In banana, the first syllable ba is weak, while the middle syllable na carries the stress. Around age two, some children pronounce banana more like “nana.” Rather than simply forgetting the first sound, they are often perceiving the word around its stressed syllable. The strongly heard NA functions as the center of the word, and the chunk nana is easier for them to grasp than the full form beginning with a weak syllable.
→ How Palkids Students Prepare for EikenVisualizing the stress pattern — the word banana
(adult pronunciation)
(around age 2)
The weak syllable ba is ignored, resulting in “nana.” This shows that children may treat the stressed syllable as the center of the word, making the weak syllable before it harder to process. Because many everyday English words follow a strong + weak trochaic pattern, children learn to use strong syllables as cues to where words begin.
Sound co-occurrence probabilities
Sound combinations that appear within the same word occur more often than combinations that cross word boundaries. Children can use this statistical pattern as a cue.
Trochaic stress pattern
Many common English words follow a strong + weak pattern (baby, table, mother, etc.). The strong syllable becomes a signal that a word is beginning.
Rules of syllable structure
When children know the sound sequences allowed in a language (phonological rules), it becomes easier to infer likely word boundaries.
The sound chunks children find are stored in the lexicon
A sound chunk heard once does not immediately become a stable word. But after repeated encounters with similar sound forms, children retain a memory that says, “I’ve heard this sound before.” If they hear the sound form dog again and again, the sequence of sounds itself stays in memory. At this stage, it does not yet need to be tied to the clear meaning “dog.”
The system that stores sound information recognized as language is called the lexicon.
The important point here is to separate storage in the lexicon from the acquisition of meaning. First, the child creates an index for “this recurring sound chunk,” and only afterward does meaning begin to develop.
Finding candidate word forms
A recurring sound chunk is segmented out of continuous speech as a likely word-like unit.
The sound chunk /dɔg/Storing it in the lexicon
A recurring sound form is stored in the lexicon as a familiar word form. At this point, the meaning can still be blank.
dog = meaning not yet assignedConnecting it with experience
As children hear “dog” while looking at real dogs, the sound form becomes linked to meaning and mental imagery.
dog = image of a dogThe earliest lexicon is like a list of headwords without meanings
In a paper dictionary, each headword comes with a definition. But when we think about a child’s early lexicon, it is easier to imagine the headword being created first, with the meaning added later.
For example, a child may hear the sound giraffe many times without ever having seen a giraffe or knowing what the word means. Even so, if the same sound form is repeated often enough, the child can still build the memory, “I hear this sound chunk again and again.” In other words, a word can exist in the mind before its meaning is understood.
Adults experience the same thing. When we repeatedly hear an unfamiliar foreign word or technical term, we often feel, “I don’t know what it means, but I’ve heard that word before.” That is an example of the sound form being remembered before the meaning.
A completed dictionary entry (adult)
- It has a headword (sound form and spelling)
- It has a clear definition
- Multiple meanings and uses are organized
- Related words and example sentences are connected
- Sound form and meaning are firmly integrated
Early lexicon (child’s beginning stage)
- The sound form (headword) is stored first
- The meaning field is blank or still vague
- Only a sense of “I’ve heard this before” exists
- Meaning becomes attached later through experience
- Sound form comes first, and meaning develops later
As word forms accumulate, new words become easier to find
As sound forms build up in the lexicon, speech segmentation itself becomes easier. If the sound form dog appears in a stream of speech and it is already stored, the child can recognize, “This part is dog.” That, in turn, makes it easier to detect the boundaries of the unknown sounds around it.
At first, language may sound like nothing more than an unbroken flow of sound. Gradually, however, children begin to hear it as a mix of “sound chunks I know” and “sound chunks I don’t know yet.” What is developing at this stage is a growing stock of word forms that children can recognize as the same word when they hear it, even before they know the meaning. That stored knowledge becomes the foundation for later vocabulary understanding.
Speech segmentation and lexicon building are not one-way processes. Instead, a cycle continues: segment → store → use stored word forms to segment further → discover new word forms. Little by little, the child’s stock of recognizable words grows.
“I’ve heard that before” becomes preparation for meaning acquisition
When a sound form has already been stored in the lexicon, it becomes easier to connect that form to meaning when the word later appears in a concrete situation. Imagine a child who has already heard the sound form dog many times and has it stored in the lexicon, then hears Look at the dog. while someone points to a real dog.
That may be the first moment when the sound form dog becomes linked to the animal in front of the child. Each new encounter with another dog, a dog in a picture book, or a running dog enriches that meaning even more.
Language development begins with finding sound chunks in continuous speech, storing those chunks in the lexicon, and then gradually building meaning on top of them.
Home English and Palkids
How to apply this process at home
What this page has described is the general language-acquisition process in which children find word-like sound chunks in continuous speech and first register them in the lexicon as sound forms without meaning. To learn more about the idea of creating a home environment rich in English audio, see What Is Home English?. For a concrete home-learning framework, see the Palkids Input Method.
