The /t/ and /d/ sounds are made in exactly the same place, with the tongue tip pressed to the bumpy ridge just behind the top front teeth, and they differ only in voicing. Say two, then do: your tongue never moves, but one is quiet and one buzzes. That single fact explains most of what goes wrong with these sounds, and it tells you how to fix each error.
This guide shows SLPs and parents how to teach /t/ and /d/ and how to sort the two error families that affect them — backing (t→k, d→g) and voicing (t↔d) — with tongue-tip placement, word lists by position, minimal pairs for each error, developmental norms, and sample IEP goals.
Errors here cost a lot of intelligibility, because these sounds are everywhere. In a classic frequency count of conversational American English, /n/ and /t/ were the two most frequent consonants (Mines, Hanson & Shoup, 1978).
The guide is written primarily for SLPs, with a clearly marked home-practice section for parents. If you are a parent who searched because your child says “key” for tea or “goor” for door, the short answer is that this pattern, called backing, is not a wait-and-see. The backing section below explains why, and why “gog” for dog can be the exception.
The Two Alveolar Stops: Quiet /t/ and Buzzy /d/
The tongue tip does most of the work in English: consonants made with the tongue tip at or just behind the top teeth make up a large share of all consonants in conversational speech. /t/ and /d/ are the two alveolar stops in that family. The tongue tip seals against the ridge, air pressure builds behind it, and the seal pops open.
They differ only in voicing:
- Voiceless /t/ — the quiet one. Vocal cords off, with a puff of air on release. Top, tooth, hat.
- Voiced /d/ — the buzzy one. Vocal cords on as the tongue lets go. Door, doll, bed.
“Quiet” versus “buzzy” is the same child-friendly handle our TH guide uses, and it runs through placement, home practice and minimal pairs here. In Shriberg’s classic grouping of English consonants, /d/ is an Early-8 sound and /t/ a Middle-8 sound, the same split you will see in the norms below. That gap is not about a harder tongue position, because the tongue does exactly the same job for both.
Both sounds also appear at the start of words, in the middle, and at the end, so one error pattern reaches a large share of a child’s vocabulary.
One American-English quirk to carry into the word lists. Between two vowels, before an unstressed syllable, /t/ and /d/ both become a quick tap called a flap: adults say water, butter, and little with a d-like sound, and that is correct pronunciation, not a voicing error. A crisp, puffed /t/ in the middle of a word lives where the /t/ starts a stressed syllable, as in hotel, potato, and guitar.
For a full single-sound reference on each phoneme, see the sound pages for the /t/ sound and the /d/ sound.
When Should a Child Say the T and D Sounds?
By 3;0, 90% of U.S. children produce /d/; by 4;0, 90% produce /t/, and about half produce both by 2;0. The values come from Crowe & McLeod’s 2020 U.S. review of 15 studies covering 18,907 children, the current standard for American English consonant acquisition, with its cross-linguistic companion, McLeod & Crowe (2018).
| Sound | 50% (customary) | 90% (mastery) | Why it is here |
|---|---|---|---|
| /d/ (voiced) | 2;0 | 3;0 | one of the earliest-mastered English consonants |
| /t/ (voiceless) | 2;0 | 4;0 | mastered about a year after /d/ |
| /k/ | 2;0 | 4;0 | for comparison — what backing swaps in for /t/ |
| /g/ | 2;0 | 4;0 | for comparison — what backing swaps in for /d/ |
Ages from Crowe & McLeod (2020); companion cross-linguistic data in McLeod & Crowe (2018).
Ages use the standard speech-pathology years;months format, not decimals. 3;6 means 3 years, 6 months old — not 3.6 years. 4;0 means 4 years, 0 months.
In the review’s own bands, /d/ belongs with /p/, /b/, /m/, /n/, /h/ and /w/, the group acquired before age 3, while /t/ belongs with /k/, /g/, /ng/, /f/ and /y/, the group acquired before age 4 (90% criterion). That matches our speech sound milestones by age chart, where /d/ is one of the “early seven” and /t/ arrives with the velars.
That frames the IEP decision for the common case. A child past 3;0 who still cannot produce /d/, or past 4;0 without /t/, with an intelligibility or participation impact, is typically a reasonable target.
Two caveats keep this honest, as with any norms-based decision: being under age does not rule out an IEP (a highly unintelligible younger child may still need service), and being over age does not mandate one (a mild, fully intelligible error may not meet eligibility).
Unlike most sounds, though, age is not the whole story for /t/ and /d/. The type of error decides how urgent it is: a voicing slip at 2;6 and backing at 2;6 are different conversations, as the next two sections explain.
Backing vs Voicing: Triage the Error
Nearly every /t/ and /d/ error falls into one of two families, and they carry very different weight. Voicing errors, where the voice switches on or off at the wrong moment, are typically gone by about 3;0 in Caroline Bowen’s elimination table. Backing, where the back of the tongue stands in for the tip, has no typical age at all.
| Pattern | Sounds like | What it is | Typical until | Fix with |
|---|---|---|---|---|
| Backing (t→k) | “key” for tea, “koy” for toy | a front sound made at the back | no typical age — atypical at any age | T-vs-K pairs |
| Backing (d→g) | “goor” for door, “goll” for doll | a front sound made at the back | no typical age — atypical at any age | D-vs-G pairs |
| Velar assimilation | “gog” for dog, “guck” for duck | a back sound later in the word pulls the first sound back | about 3;0 (ASHA) | often fades; probe velar-free words first |
| Prevocalic voicing (t→d) | “dop” for top, “do” for two | the voice switches on too early | about 3;0 (Bowen) | T-vs-D pairs, /t/ word as the target |
| Final devoicing (d→t) | “bet” for bed, “beet” for bead | the voice cuts out too soon | about 3;0 (Bowen) | T-vs-D pairs, /d/ word as the target |
| Stopping (s→t) | “tun” for sun | /t/ is the substitute, not the target | varies by sound | fricative work, not /t/ work |
Backing is the one to take seriously, and its tell is that it shows up in words with no back sound anywhere: tea, toy, door, doll. Velar assimilation is the look-alike: dog ends in /g/ and duck in /k/, so a young child may copy the back sound to the front of the word. ASHA’s table of typical phonological patterns lists it, with “guck” for duck as its example and a likely elimination age of about 3.
Voicing errors keep the placement and miss the switch, so they need no placement work. In stopping (“tun” for sun) the child’s /t/ is fine, because it is the substitute; the work belongs to the fricative, as our phonological processes guide explains.
Three quick calls at the end of the word, where /t/ and /d/ are easiest to mis-score:
- An unreleased final /t/ is normal. Adults often finish hat or boat with the tongue held on the ridge and no puff of air. Score it correct.
- Final-consonant deletion is typical early. Dropping the sound altogether (“ca” for cat) usually fades by about 3, per ASHA’s table.
- A throat catch can go either way. A glottal stop for /t/ in button or kitten is ordinary American speech, while glottal stops replacing many different consonants belong with the atypical patterns and are worth an SLP look.
Backing can reach beyond /t/ and /d/, too: /n/ can become “ng” (sun as “sung”) and /s/ can slide back toward “sh,” and Sound Safari’s error-pattern labels count both as backing. On the app’s screener the labels follow the table: a /t/ marked as “k” or a /d/ marked as “g” counts as backing, a /t/ marked as “d” as voicing, and a /d/ marked as “t” as devoicing.
Each row maps to a minimal-pair set below: voicing errors to T-vs-D pairs, backing to T-vs-K and D-vs-G pairs.
Why Backing Is Different From Fronting (Evaluate, Don’t Wait)
Fronting and backing are mirror images, but only one is a normal stage. Fronting (cat as “tat,” go as “doe”) is a typical process that usually fades by about 3;6 to 4;0. Backing (tea as “key,” door as “goor”) appears on neither ASHA’s list of typical patterns nor Bowen’s elimination table, so there is no age at which it is expected to fade.
ASHA’s guidance is direct about what that means: patterns that persist, or that are atypical, “lead to a diagnosis of phonological disorders.”
In Barbara Dodd’s widely used model of speech sound disorders, backing is a signature example of consistent phonological disorder, the consistent use of unusual error patterns that typical development does not produce (Dodd, 2014), and her group’s trials point to phonological contrast therapy for it. An Australian community study of 4-year-olds put that group at 20% of the 126 children with a speech sound disorder, against 55% with ordinary phonological delay (Ttofari Eecen et al., 2019).
You may read elsewhere that backing is common and natural until age 5, and that you should evaluate only after that. The primary sources say otherwise. Our /k/ and /g/ guides, which cover the mirror error, take the same line: backing is atypical at any age and warrants an evaluation whenever it shows up as a pattern.
Two checks keep the call accurate:
- Rule out assimilation. If the only backed words already contain /k/ or /g/ (dog, duck), probe words with no back sound in them (door, doll, tea, toy). Backing shows up in those; assimilation does not.
- Rule out dialect and home language. A pattern that matches the child’s home dialect or first language is a difference, not an error, as ASHA’s patterns guidance makes explicit.
Unusual patterns are also one reason an SLP or pediatrician may suggest a hearing check, and it is a reasonable thing to ask about.
Sound Safari is a clinical tool, not a medical device. It does not provide diagnoses or treatment recommendations. Clinical decisions should always be made by qualified professionals, including the call on whether a pattern is backing, assimilation or dialect.
The typical processes that do fade on schedule, fronting and assimilation among them, are covered in the phonological processes guide linked above.
How to Teach the T and D Sounds: Tongue-Tip Placement
Because /t/ and /d/ share one placement, the five steps below teach the tongue position once and split the two sounds with a single voicing switch. A child who backs needs the placement and listening work in steps 2 and 5; a child with a voicing error can usually move quickly to the switch in steps 3 and 4.
Step 1 — Probe stimulability and sort the error
Model /t/ and /d/ on their own, then in a matched pair such as two and do, and listen for two things: which sound is closer, and what the child says instead. A child who backs needs placement work first, while a child with a voicing error already has the placement and needs only the switch. If one sound is usable, start there; a steady /d/ gets you /t/ cheaply, and the reverse is also true.
While you probe, include a few words with no back sound in them, such as toy, tea, door and doll, so assimilation cannot masquerade as backing.
Step 2 — Find the bumpy spot
Show the child the alveolar ridge, the bumpy spot just behind the top front teeth, and have the tongue tip touch it. A mirror lets the child see the tip go up and stay at the front, and a small flashlight helps. For a child who backs, touch the ridge first with a clean finger or a tongue depressor, then ask the tongue tip to find the same spot.
The NHS Wirral home programme for children who say “g” for “d” uses the same cue: tongue behind the top teeth, tip kept at the front. For a child who backs, add a front-versus-back label: point to the front teeth for /t/ and /d/ and to the throat for /k/ and /g/.
This guide skips food cues such as a dab of peanut butter on the ridge, since a clean finger does the same job without allergy questions. It also skips tongue-strengthening drills: a systematic review in ASHA’s American Journal of Speech-Language Pathology found insufficient evidence that non-speech oral-motor exercises improve speech (McCauley et al., 2009).
Step 3 — Build the quiet /t/
With the tongue tip on the ridge, let air pressure build, then release it in a quick, crisp puff. Hold a tissue or a strip of paper in front of the lips: at the start of a word, /t/ makes it flutter and /d/ barely moves it. A clock-tick cue (t-t-t) keeps the release short and sharp.
Practice it alone, then with vowels, as in “tee” and “too.” If the release slides back toward /k/, return to step 2 before going on.
Step 4 — Add the voice for /d/
Keep the same tongue position and turn the voice on. With a hand on the throat, a slow, voiced “duh” buzzes, while a whispered “t” stays quiet until the vowel starts. Then toggle: whisper the start of a word to switch the voice off, and say it aloud to switch it back on; for a voicing error, this toggle is the therapy target.
For prevocalic voicing (“dop” for top), work from the quiet side: whisper top, then say it aloud and keep the puff. For final devoicing (“bet” for bed), work from the buzzy side and stretch the vowel. English speakers lengthen a vowel before a voiced /d/, so the longer vowel in bead than in beet is half the cue.
Step 5 — Watch for drift, then generalize
Catch two drifts early: the tongue sliding back toward the soft palate, which drifts toward /k/ and /g/, and the tongue tip poking out between the teeth, which drifts toward TH. Re-anchor at the ridge with the mirror, and for a child who backs, put listening before talking by sorting pictures into front and back sounds first. Then climb the practice hierarchy: isolation, syllable, word, phrase, sentence, story.
That listening-first order is how the NHS Wirral backing programme is built. The child hears “d” and “g” apart, sorts pictures by their first sound, and tells rhyming words apart before being asked to say /d/ at all, in sessions of about 15 minutes at least three times a week.
Sound Safari’s six practice levels, Isolation → Syllable → Word → Phrase → Sentence → Story, line up with the word lists and goals below, with ready-made /t/ and /d/ phrases and sentences for every word position.
T Word Lists by Position
Most therapy starts with word-initial /t/, the easiest position to cue, moves to the end of the word, where an unreleased /t/ counts as correct, and finishes with medial. Sound Safari ships 89 /t/ practice words: 30 initial, 29 medial, and 30 final. The initial and final samples below are free of /k/ and /g/, so a child who backs cannot borrow a back sound from elsewhere in the word.
| Initial /t/ | Medial /t/ | Final /t/ |
|---|---|---|
| toy | hotel | hat |
| tooth | potato | boat |
| top | guitar | feet |
| table | cartoon | heart |
| turtle | water | nut |
Medial /t/ needs the most care. All but four of the 29 medial /t/ words in the app’s list are flapped or glottalized in everyday American speech: water, butter, and little get the quick d-like flap, while kitten, button, and mitten get a catch in the throat. Both are correct adult pronunciation, so do not score a d-like water as a voicing error.
The crisp four in the table, hotel, potato, guitar and cartoon, start a stressed syllable with /t/, and water is there as the flap example. Guitar and cartoon carry a back sound, so save them until /t/ is steady.
Four of the 30 initial /t/ words — tree, truck, train, and tractor — start with a TR- blend rather than a singleton /t/. Even adults pull that /t/ back toward the sound at the start of chew (tree can sound close to “chree”), so keep these words out of singleton work; they belong with the blend work in our cluster reduction guide.
Sound Safari sorts the full /t/ list by position for you, so you are picking from a set rather than building one by hand.
D Word Lists by Position
Sound Safari ships 90 /d/ practice words: 30 initial, 30 medial, and 30 final. As with /t/, the initial and final samples avoid /k/ and /g/.
| Initial /d/ | Medial /d/ | Final /d/ |
|---|---|---|
| door | spider | bed |
| doll | ladder | red |
| dinosaur | panda | bead |
| daisy | window | seed |
| dish | noodle | food |
With a child who backs, save words that carry a back sound, such as dog, duck, doctor, donkey, dig and desk, for later, since the back sound can pull the /d/ along with it. Seven of the 30 initial /d/ words — dragon, dress, drink, drive, drop, drum, and dry — are DR- words, where even adults move /d/ toward the sound at the start of jump (drum can sound close to “jrum”); they belong with the R-blend work linked above.
Final /d/ is where devoicing shows up (“bet” for bed), so pair the buzz with a stretched vowel, as in step 4. Medial /d/, between two vowels, is the easiest place to keep the voice running, which makes it a good bridge into the voicing pairs below. Both full lists live on the /t/ and /d/ sound pages linked earlier.
T vs D Minimal Pairs: Fixing Voicing Errors
A voicing error is the easiest /t/ and /d/ problem to treat with minimal pairs, because the placement is already right and only the switch is wrong. Caroline Bowen’s point holds here as it does for other sounds: as few as 3–5 well-chosen minimal pairs can drive generalization to untreated words (Elbert et al., 1991, as Bowen cites it).
Sound Safari’s curated T-vs-D set, filed under voicing in the app’s Minimal Pairs game, carries seven pairs:
| T | D |
|---|---|
| ten | den |
| two | do |
| time | dime |
| town | down |
| tip | dip |
| bat | bad |
| hat | had |
At the time of writing, the app also ships 26 illustrated d–t pairs, including picture versions of several pairs above and voiced-ending pairs such as bead/beet and seed/seat, so you can run the contrast without printing worksheets.
The same deck treats both voicing errors; you flip which word is the target. For prevocalic voicing (“dop” for top), the target is the /t/ word: ten, not den. For final devoicing (“bet” for bed), the target is the /d/ word: bad, not bat.
Why the contrast works: hearing time versus dime makes the voicing switch meaningful, because the two words mean different things. Browse more T/D pairs at T vs D minimal pairs.
T vs K and D vs G Minimal Pairs: Fixing Backing
Backing is treated with front-versus-back contrast pairs, the phonological contrast approach that Dodd’s group’s trials point to for consistent atypical patterns like backing. The pairs make the front placement meaningful, because tea and key are two different things.
One labeling note, so the app does not confuse you: Sound Safari files its T-vs-K and D-vs-G decks under velar fronting, the far more common error. They serve backing just as well once you flip the target word. For fronting (“tea” for key), the target is the /k/ word; for backing (“key” for tea), the target is the /t/ word.
At the time of writing, the app ships 38 illustrated k–t pairs and 13 d–g pairs. A working sample:
| T vs K (backing: target the /t/ word) | D vs G (backing: target the /d/ word) |
|---|---|
| tea / key | dot / got |
| tap / cap | down / gown |
| tub / cub | bed / beg |
| toast / coast | mud / mug |
Tea and toast are also in the app’s /t/ practice list, and bed and mud in its /d/ list, so you can carry them straight into word drills.
Run the listening step before any production, as in step 5: the child sorts pictures by whether they start with a front sound or a back sound, using the step 2 cue of front teeth for /t/ and /d/ and throat for /k/ and /g/. Then show both pictures, ask for one, and respond to whatever the child actually says. If they back tea to “key,” hand over the key picture; the mismatch does the teaching.
Browse the K vs T minimal pairs or the full minimal pairs library. The fronting side of these same decks is covered in the /k/ and /g/ guides linked above.
Sample IEP Goals for /t/ and /d/
Once /t/ or /d/ is confirmed as a target (past 3;0 for /d/ or 4;0 for /t/, with an intelligibility impact, or backing at any age), write sound goals for articulation errors and one pattern goal for backing or voicing. Sound Safari’s goal bank ships 22 built-in templates across 8 categories (articulation, phonology, fluency, voice, receptive language, expressive language, pragmatics, and phonological awareness), all defaulting to an 80% accuracy criterion across 3 consecutive sessions. (The sample goals below say “20 trials in 3 of 4 consecutive sessions” — the “20 trials” and “3 of 4” are common clinician conventions layered on the app’s underlying 3-consecutive-session default, not app values.)
For a child who cannot yet produce /t/ or /d/ correctly, the articulation “Sound in Word Position” template fits: suggested baseline 0–40%, 80% accuracy, 12–36 weeks.
Goal 1 (/t/, word position): Given a verbal model and tactile cues, [Student] will produce /t/ in the initial position of single-syllable words (e.g., toy, tooth, top) with 80% accuracy across 20 trials in 3 of 4 consecutive sessions by the end of the IEP year. Suggested baseline: 0–40%. Timeframe: 12–36 weeks.
Goal 2 (/d/, word position): Given a verbal model and tactile cues, [Student] will produce /d/ in the initial position of single-syllable words (e.g., door, doll, dish) with 80% accuracy across 20 trials in 3 of 4 consecutive sessions by the end of the IEP year. Suggested baseline: 0–40%. Timeframe: 12–36 weeks.
The app has no isolation-level template, so for a child who is not yet stimulable, the goal below follows the isolation goals in our IEP article; its values are illustrative, not app values:
Goal 3 (isolation — illustrative): Given a verbal model and tactile cues (tongue tip to the ridge behind the top teeth), [Student] will produce /t/ and /d/ in isolation with 85% accuracy across 20 trials in 3 of 4 consecutive sessions. Suggested baseline: 0–30%. Timeframe: 6–10 weeks.
Backing and voicing errors are different: they are patterns, and backing usually takes /t/ and /d/ together, so each gets one phonology-style goal instead of two parallel sound goals. The app has phonology templates for fronting and stopping but no backing or voicing template, so both goals below are written by hand; the backing goal follows the shape of the app’s fronting template, and all values are illustrative, not app values:
Goal 4 (backing, phonology — illustrative): [Student] will eliminate the process of backing by producing the alveolar sounds /t/ and /d/ in place of velars in the initial position of words with 80% accuracy across 3 consecutive sessions by the end of the IEP year. Illustrative baseline: 0–30%. Illustrative timeframe: 20–40 weeks.
Goal 5 (voicing contrast — illustrative): Given picture pairs and a verbal model, [Student] will produce /t/ and /d/ correctly in minimal-pair words that differ only in voicing (e.g., ten/den, bat/bad) with 80% accuracy across 20 trials in 3 of 4 consecutive sessions. Illustrative baseline: 20–40%. Illustrative timeframe: 12–20 weeks.
The app operationalizes these goals with two scaffolds. Trials are scored Correct / Approximate / Incorrect, and Approximate is an opt-in setting, so the default is simply Correct/Incorrect. There are also four prompting levels, Maximum (Picture + Word + Model), Moderate (Picture + Word), Minimal (Word Only), and Independent (Picture Only), and fading from Maximum to Independent is how the goal’s cue condition tightens over the year.
Every goal here is a starting point, not a substitute for clinical judgment, baseline data, or your district’s conventions. For the goal formula, the fronting goal the backing goal mirrors, and more sound-by-sound templates, see our IEP goal templates. In Sound Safari, session scoring links straight to the goals you have written, so the probe data behind each one stays current without manual re-entry.
Home Practice for Parents
This section is for parents and caregivers, and no clinical background is needed. (An “SLP” is a speech-language pathologist, the professional who treats speech sounds.)
First, reassurance. Mixing up the quiet t and the buzzy d, as in “dop” for top or “bet” for bed, is common at age 2 and usually fades by about age 3.
Now the one honest flag. When back sounds replace front ones, so tea comes out “key” and door comes out “goor,” that is not a normal stage; if you hear it consistently, at any age, ask an SLP to take a look, and ask about a hearing check too. That is not a reason to panic, but it is a reason not to wait.
One thing to check first: if the only words affected are ones like dog (“gog”), your child may simply be copying the g at the end of the word, a habit that typically fades by about age 3. Listen for door, doll, toy, and tea, which have no back sound to copy.
A few simple things help at home:
- Find the bumpy spot. Have your child run a clean finger along the ridge just behind the top front teeth, then touch the same spot with the tongue tip. A mirror turns it into a game.
- Quiet vs buzzy. Hold a tissue in front of your child’s mouth and say top: the quiet t makes it flutter. Then put a hand on the throat for a slow, buzzy d and feel the motor run.
- Front-or-back sorting. Gather pictures of things that start with a front sound (tea, toy, door) and a back sound (key, cow, goat), say each word, and let your child sort them into two piles. Listening comes before saying.
- Model, don’t correct. If your child says “goor” for door, skip the redo and say it back naturally two or three times: “Yes, the door. Open the door!”
- Short and frequent. A few minutes, a few times a week, beats one long session. One NHS home programme for children who say “g” for “d” suggests about 15 minutes, at least three times a week.
Skip the peanut butter and the tongue exercises. The bumpy spot is easy to find with a clean finger, and an expert review found too little evidence that tongue-strengthening or blowing exercises improve speech sounds.
If your child’s SLP uses Sound Safari, they can send you a homework link with T and D practice that opens right in your web browser, no app to download. On privacy: no third party ever receives your child’s clinical or student data. No ads, no behavioral tracking, no data brokers. Account sign-in and subscriptions are handled by standard providers that see only account and billing details. See what parents get.
Sound Safari is a clinical tool, not a medical device. It does not provide diagnoses or treatment recommendations. An SLP is the right person to say whether your child’s pattern needs therapy.
From Practice to Progress: Tracking T and D Therapy
The hardest part of /t/ and /d/ therapy is knowing which error family a child is producing, and Sound Safari’s built-in screener, covering 24 sounds across initial, medial, and final positions, helps you sort the error family. When you mark what the child said in place of /t/ or /d/, the automatic error-pattern detection files a “k” for /t/ or a “g” for /d/ as backing, badged Atypical at any age (an informational flag for the clinician, not a diagnosis), and a “d” for /t/ or a “t” for /d/ as voicing errors. Together with your own probe, that points you to the T-vs-D deck or the T-vs-K and D-vs-G decks. One difference from the triage table: the screener keeps a devoicing pattern in its developing column until 4;0, a more conservative window than Bowen’s 3;0.
The screener samples /t/ with top, guitar, and cat, and /d/ with dog, panda, and bed, checking each sound against developmental norms. If dog comes out “gog,” the screener’s alternative words, door and doll, give you a quick velar-free check before you call it backing. The screener is part of Pro, so there’s no separate assessment app to switch to.
And it generates SOAP notes inside the articulation app: session data flows into the standard four-section clinical structure (Subjective, Objective, Assessment, Plan), so the documentation is largely written by the time the session ends. That is the whole workflow — practice, screen, document, and share, all in one app.
For the screening side, see our articulation screening checklist; for the documentation side, see writing SOAP notes in speech therapy.
Sound Safari is a clinical tool, not a medical device. It does not provide diagnoses or treatment recommendations. The screener is informational, and clinical decisions should always be made by qualified professionals.
To put the full loop to work, Sound Safari for professionals was built for exactly this.
Frequently Asked Questions
At what age should a child say the T and D sounds?
About half of children produce both /t/ and /d/ by age 2 (2;0), per Crowe & McLeod (2020). At the 90% mastery level, /d/ comes first, by about age 3 (3;0), and /t/ follows by about age 4 (4;0). Errors before those ages are often typical, with one exception: backing, where /t/ and /d/ are replaced by /k/ and /g/ (“key” for tea), is not a normal stage at any age.
Is it normal for a child to say “k” for “t”?
Not as a normal stage. Saying k for t (“key” for tea, “kop” for top) is called backing; it is absent from ASHA’s list of typical phonological patterns, and it is a classic example of the atypical patterns in Barbara Dodd’s model of speech sound disorders. A consistent k-for-t pattern is worth an SLP evaluation at any age, though if the word already contains a k or g sound, the child may simply be copying it, a habit that usually fades by about age 3. Sound Safari is a clinical tool, not a medical device. It does not provide diagnoses or treatment recommendations.
What is backing in speech?
Backing is a phonological pattern in which sounds made at the front of the mouth are replaced by sounds made at the back: /t/ becomes /k/ (“key” for tea) and /d/ becomes /g/ (“goor” for door). It is the mirror image of fronting (“tat” for cat), which is typical and usually fades by about 3;6 to 4;0, while backing is atypical at any age. It is usually treated with front-versus-back minimal pairs such as tea/key and dot/got.
Why does my child say “gog” for dog?
Two reasons are common. Dog ends in a back sound, g, and young children often copy a sound from later in a word; this velar assimilation is on ASHA’s list of typical patterns and usually fades by about age 3. If your child also says “goor” for door or “key” for tea, words with no back sound to copy, that is backing, which is not a typical stage and calls for an SLP evaluation rather than a wait-and-see. It is also reasonable to ask about a hearing check. Clinical decisions should always be made by qualified professionals.
What is the difference between the T and D sounds?
They are made in exactly the same place, with the tongue tip pressed to the bumpy ridge behind the top front teeth, and differ only in voicing. /t/ is the quiet sound: vocal cords off, with a puff of air that makes a tissue flutter at the start of a word. /d/ is the buzzy sound: vocal cords on, which you can feel with a hand on the throat. Pairs like two/do and time/dime differ only by that switch.
How do you teach the T sound?
Start by finding the alveolar ridge, the bumpy spot just behind the top front teeth, with a mirror and a clean finger. Have the child lift the tongue tip to that spot, build up air, and release it in a quick puff; a tissue held in front of the lips should flutter. Practice the sound alone, then in syllables like “tee” and “too,” then in words. Once /t/ is steady, add the voice to get /d/.
When should we see a speech therapist about T or D?
See an SLP if you hear backing (“key” for tea, “goor” for door) consistently at any age, if voicing mix-ups (“dop” for top, “bet” for bed) continue past about age 3, if /d/ is still missing past about age 3 or /t/ past about age 4, or if unfamiliar listeners struggle to understand your child. A free school-based screening or a private evaluation can tell you whether therapy is needed.
Closing
Teach /t/ and /d/ by remembering what they are: one tongue-tip placement with two voices, the quiet /t/ and the buzzy /d/. Sort the error before you choose the fix: voicing errors go to T-vs-D pairs, and backing goes to T-vs-K and D-vs-G pairs. And keep the series’ rule straight: fronting waits, backing gets evaluated.
The goals and templates here are starting points, not a replacement for your clinical judgment and baseline data. Use them as scaffolding and adjust to the student in front of you.
If you want structured /t/ and /d/ practice with position-sorted word lists for both sounds, labeled minimal-pair sets, automatic error-pattern detection, and progress an SLP can review, Sound Safari was built for this workflow. 14-day free trial. Cancel anytime through Apple.