Phonological processes explained: patterns in children's speech and when they stop

Phonological processes are predictable patterns that young children use to simplify adult speech, such as leaving off the end of words (“ca” for “cat”) or replacing k with t (“tat”). They are a normal part of learning to talk, and most disappear on a fairly predictable timetable between about 2 and a half and 6. A pattern that continues well past its usual age, or one that is not part of typical development, is a reason to ask a registered speech-language pathologist.

Why children simplify speech

Adult speech is full of sounds and sequences that need precise, fast movements. A toddler learning to talk has a limited set of sounds, and a limited ability to string them together. Rather than wait until every sound is ready, children talk anyway, using simpler versions of the words they hear. The simplifications are not random. The same handful of patterns show up in children all over the world, which is why clinicians can name them and give each one an approximate age.

The key word is pattern. A child who says “tat” for “cat”, “doat” for “goat” and “tea” for “key” is not making three separate mistakes. They are applying one rule: make back-of-the-mouth sounds at the front. That has practical consequences for therapy, because treating the rule tends to fix many words at once.

The common processes and when they usually stop

The ages below are approximate ages by which each pattern has usually disappeared, drawn from the summary table published by Caroline Bowen (2011) on speech-language-therapy.com, which is itself based on norms such as Grunwell (1987). They describe when a pattern is typically gone, not the last possible moment. Different sources vary by a few months.

A note on “2;6” style ages: the number before the semicolon is years and the number after it is months, so 3;3 is three years and three months.

These ages line up well with the sound-by-sound data in Crowe and McLeod (2020). For instance, k and g reach the 90 percent criterion at around 3;2 and 3;1, which is why fronting is expected to fade by about 3;6, and r reaches it at around 5;7, which is why gliding lingers longest. Our speech sound development chart gives the full list of 24 consonants.

Two further typical patterns are worth knowing. Voicing changes (saying “big” for “pig” or “back” for “bag”) are common until about 3. Deaffrication (saying “ship” for “chip”) usually fades by around 4. Both are listed on Bowen’s table as developmental.

Patterns that are not typical

Some patterns are not a normal step on the way to adult speech at any age. The ASHA Practice Portal and Bowen’s summary both describe these as atypical or non-developmental. They include:

• Backing. Front sounds made at the back: “key” for tea, “gog” for dog when there is no g in the word to copy.
• Initial consonant deletion. Leaving off the first sound: “at” for cat, “og” for dog.
• Glottal replacement. Replacing a consonant with a catch in the throat: “bo-uh” for bottle.
• Vowel errors after about age 3.
• Lateral lisp. Air escaping over the sides of the tongue on s, z or sh (see our guide to lisps).
• Unusual or inconsistent substitutions that change from one attempt to the next.

None of these means something is seriously wrong on its own. They simply do not fit the usual developmental map, so an SLP will want to look more closely. Inconsistent errors in particular are one of the features clinicians consider when they think about motor planning, which is covered in our guide to childhood apraxia of speech.

Delay, disorder, and the difference from articulation

Clinicians use a few terms that are worth separating.

• Phonological delay means the child is using typical processes, but past the usual age. A 4-year-old who still fronts k to t fits this description. Delays are the most common picture and often respond well to therapy.
• Phonological disorder usually means atypical patterns are present, or typical patterns are unusually persistent and widespread.
• Articulation difficulty is a motor problem with producing a specific sound, such as a frontal lisp. The sound comes out distorted rather than being swapped by rule. Articulation errors are less tied to age.

The ASHA Practice Portal groups all of these under the umbrella of speech sound disorders, and many children show a mix. The distinction matters because the therapy approaches differ. Articulation therapy teaches the mouth to make a sound. Phonological therapy teaches the child that the sound matters, so they start using one they can already make.

How minimal pairs therapy works

The most widely known phonological approach is minimal pairs. A minimal pair is two words that differ by a single sound: key and tea, boat and bow, sun and ton, spoon and poon (a made-up word is fine). Each pair is chosen so that the child’s error turns one word into the other.

A typical session, as described in the ASHA Practice Portal and in Bowen’s clinical writing, goes something like this.

1. Listening first. The clinician shows pictures of both words and says them. The child points. This builds the awareness that “tea” and “key” are different words.
2. Production with a purpose. The child asks for a picture by name. If a child who fronts says “tea” while pointing at the key, the clinician hands them the tea. The mismatch is the lesson: the sound changed the meaning. The clinician does not scold; they simply respond to what was said and give another chance.
3. Lots of trials, lots of pairs. Because the goal is to change a rule, therapy uses many different word pairs that all involve the same contrast, rather than drilling one word.
4. Generalization. Once the contrast is used reliably in the pairs, it is carried into sentences and conversation. Often words that were never practised improve too, because the underlying pattern has shifted.

Related approaches include multiple oppositions (several contrasts at once for children with many errors), maximal oppositions (contrasting very different sounds), and the cycles approach, which rotates through patterns in short blocks. An SLP chooses based on how many patterns a child has, their age and their stimulability.

Our free minimal pairs tool generates picture pairs for the common processes, which can be useful for the listening games above once an SLP has confirmed which pattern to target.

What you can do at home

• Listen for the pattern, not the word. If you notice several words with the same change, such as all the k words coming out as t, write a few down. It is useful information for an assessment, and the speech sound checker can help you organize it.
• Model the full word. When your child says “poon”, respond with “Yes, your spoon. A big spoon.” Stretching the missing sound slightly helps it stand out without a correction.
• Play listening games with minimal pairs before asking for production. “Point to the key. Now point to the tea.” Listening is a skill in its own right.
• Keep ends of words audible. Children who delete final consonants benefit from hearing them clearly, so slow down a little on words like “cup” and “hat”.
• Check hearing, especially after repeated ear infections, because fluid in the middle ear can blur exactly these contrasts.
• Ask before drilling. Practising the wrong target, or the right target in the wrong way, can slow things down. The articulation practice tool and more games are in our guide to speech therapy activities at home.

When to ask for an assessment

Consider contacting a registered speech-language pathologist if your child is using a pattern well past the age in the table (for example, still deleting final consonants at 4), is using any of the atypical patterns above, is hard for people outside the family to understand at 3, or is frustrated by not being understood. In Ontario, children who have not yet started school can be self-referred to the free Preschool Speech and Language Program with no doctor’s referral. Our guide on when to see a speech-language pathologist lists the other common referral signs. Earlier support is generally easier, and an assessment gives clarity either way.

A note for clinicians

Process ages in the table reflect the Bowen (2011) summary rather than a single normative sample, so they will not match every test manual, and families reading different charts may see figures that differ by a few months. Pairing the process table with the 90 percent consonant ages in Crowe and McLeod (2020) and the cross-linguistic data in McLeod and Crowe (2018) can help when explaining why fronting and gliding sit where they do, and when discussing patterns in children who are learning English alongside another language.