Your vowel space

Blue: where each of your vowel sets sits, and how much it varies. Grey: where the American reading pool (LibriSpeech) puts the same sets. Left is front, right is back, top is high.

Where your vowels land

• FLEECE (see) — sits with the other General American speakers here (American pool: same position)
• KIT (sit) — sits with the other General American speakers here (American pool: same position)
• DRESS (bed) — sits with the other General American speakers here (American pool: same position)
• TRAP (cat) — sits with the other General American speakers here (American pool: same position)
• STRUT (cup) — sits with the other General American speakers here (American pool: same position)
• LOT (odd) — sits with the other General American speakers here (American pool: same position)
• THOUGHT (dawn) — sits with the other General American speakers here (American pool: closer to LOT (odd), 100% of the way)
• FOOT (good) — sits with the other General American speakers here (American pool: closer to GOOSE (food), 69% of the way)
• GOOSE (food) — near the other General American speakers here (American pool: same position)

What we noticed

• You say cot and caught with the same vowel — common in the western US, Canada, and much of Ireland.
• On the word list, the better test for this pair, cot and caught come out as one vowel; the story reading suggested no clear answer. The word list stands.
• Put and putt are clearly different vowels for you.
• Your cat vowel stays put before n and m — no American-style raising. As most General American speakers here do.
• Your food vowel is fronted — typical of younger Americans and Southern England. As most General American speakers here do.
• Your day vowel barely glides — as in Northern England, Scotland, much of Ireland and the Upper Midwest. As most General American speakers here do.
• Your go vowel barely glides — as in Northern England, Scotland and much of Ireland. Unusual among the General American speakers here.
• You pronounce r after vowels (nurse, car) — a rhotic accent, as in Ireland, Scotland and North America. As most General American speakers here do.
• Your aunt rhymes with ant.
• When talking freely, your vowels move toward the centre and contrasts shrink — this happens to everyone.
• Your FOOT–GOOSE vowels sit in almost the same place but differ clearly in length — a distinction carried by duration rather than by position, which the chart cannot show.

Word list

Short words read as a list, five or so of each kind. These pin down the pairs the story could not. Verdicts here come from a label-shuffle test on the word list alone.

• cot / caught: one vowel (separation 0.39, p 0.819, 6 + 5 words)
• put / putt: two different vowels (separation 1.14, p 0.007, 3 + 5 words)
• pin / pen: two different vowels (separation 9.62, p 0.01, 4 + 4 words)
• soot / suit: too few clear tokens (3 + 0 words)
• beer / bear: too few clear tokens (0 + 3 words)
• fur / fair: two different vowels (separation 1.61, p 0.026, 4 + 3 words)
• Mary / marry: too few clear tokens (1 + 1 words)
• Mary / merry: too few clear tokens (1 + 1 words)
• marry / merry: too few clear tokens (1 + 1 words)
• poor / pour: too few clear tokens (2 + 1 words)
• horse / hoarse: too few clear tokens (0 + 1 words)
• bath / trap: close — not clearly apart (separation 1.4, p 0.277, 4 + 4 words)
• ban / bat: too few clear tokens (0 + 4 words)
• ride / write: too few clear tokens (0 + 0 words)
• loud / lout: too few clear tokens (0 + 0 words)

Reference groups

Fewer than three other speakers from your background yet — your chart is compared with the American reading pool. Distance from your vowel system to each group's; smaller is closer. The first number is by chart position; the second places you by the measured accent features (contrasts, positions, glides, r after vowels), which is still being validated. Groups with fewer than three speakers are one or two individuals, not an accent — read those rows as hints.

1. General American (8 speakers) — 0.23 by chart position · 0.64 by features
2. Irish English (Limerick) (4 speakers) — 0.37 by chart position · 0.77 by features
3. Irish English (other regions) (3 speakers) — 0.48 by chart position · 0.88 by features
4. Irish English (Dublin) (2 speakers — provisional) — 0.62 by chart position · 1.44 by features
5. Other native English (British, Australian…) (5 speakers) — 0.64 by chart position · 1.28 by features
6. Spanish L1 learners (2 speakers — provisional) — 0.69 by chart position · 1.16 by features
7. Other-L1 learners (1 speakers — provisional) — 0.75 by chart position
8. Irish English (Ulster) (3 speakers) — 0.8 by chart position · 1.46 by features

Tu espacio vocálico

Azul: dónde se sitúa cada uno de tus grupos de vocales y cuánto varía. Gris: dónde los sitúa el grupo americano de lectura (LibriSpeech). Izquierda es anterior, derecha posterior, arriba cerrada.

Dónde caen tus vocales

• FLEECE (see) — se sitúa con los demás hablantes de General American de aquí (grupo americano: misma posición)
• KIT (sit) — se sitúa con los demás hablantes de General American de aquí (grupo americano: misma posición)
• DRESS (bed) — se sitúa con los demás hablantes de General American de aquí (grupo americano: misma posición)
• TRAP (cat) — se sitúa con los demás hablantes de General American de aquí (grupo americano: misma posición)
• STRUT (cup) — se sitúa con los demás hablantes de General American de aquí (grupo americano: misma posición)
• LOT (odd) — se sitúa con los demás hablantes de General American de aquí (grupo americano: misma posición)
• THOUGHT (dawn) — se sitúa con los demás hablantes de General American de aquí (grupo americano: más cerca de LOT (odd), un 100% del camino)
• FOOT (good) — se sitúa con los demás hablantes de General American de aquí (grupo americano: más cerca de GOOSE (food), un 69% del camino)
• GOOSE (food) — cerca de los demás hablantes de General American de aquí (grupo americano: misma posición)

Lo que hemos notado

• Pronuncias cot y caught con la misma vocal — habitual en el oeste de EE. UU., Canadá y gran parte de Irlanda.
• En la lista de palabras, la mejor prueba para este par, cot y caught salen como una sola vocal; la lectura del cuento sugería una respuesta poco clara. Vale la lista.
• Put y putt son vocales claramente distintas para ti.
• Tu vocal de cat no cambia ante n y m. Como la mayoría de los hablantes de General American de aquí.
• Tu vocal de food es anterior — típico de los americanos jóvenes y del sur de Inglaterra. Como la mayoría de los hablantes de General American de aquí.
• Tu vocal de day apenas se desliza — como en el norte de Inglaterra, Escocia, gran parte de Irlanda y el Medio Oeste alto de EE. UU. Como la mayoría de los hablantes de General American de aquí.
• Tu vocal de go apenas se desliza — como en el norte de Inglaterra, Escocia y gran parte de Irlanda. Poco habitual entre los hablantes de General American de aquí.
• Pronuncias la r tras vocal (nurse, car) — un acento rótico, como en Irlanda, Escocia y Norteamérica. Como la mayoría de los hablantes de General American de aquí.
• Tu aunt rima con ant.
• Al hablar libremente, tus vocales se acercan al centro y los contrastes se reducen — le pasa a todo el mundo.
• Tus vocales de FOOT–GOOSE caen casi en el mismo sitio pero se distinguen claramente por duración — una distinción que lleva la longitud y no la posición, y que el gráfico no puede mostrar.

Lista de palabras

Palabras cortas leídas como lista, unas cinco de cada tipo. Fijan los pares que el cuento no podía. Los veredictos salen de una prueba de barajado de etiquetas sobre la lista sola.

• cot / caught: una sola vocal (separation 0.39, p 0.819, 6 + 5 words)
• put / putt: dos vocales distintas (separation 1.14, p 0.007, 3 + 5 words)
• pin / pen: dos vocales distintas (separation 9.62, p 0.01, 4 + 4 words)
• soot / suit: pocos tokens claros (3 + 0 words)
• beer / bear: pocos tokens claros (0 + 3 words)
• fur / fair: dos vocales distintas (separation 1.61, p 0.026, 4 + 3 words)
• Mary / marry: pocos tokens claros (1 + 1 words)
• Mary / merry: pocos tokens claros (1 + 1 words)
• marry / merry: pocos tokens claros (1 + 1 words)
• poor / pour: pocos tokens claros (2 + 1 words)
• horse / hoarse: pocos tokens claros (0 + 1 words)
• bath / trap: cercanas — no claramente separadas (separation 1.4, p 0.277, 4 + 4 words)
• ban / bat: pocos tokens claros (0 + 4 words)
• ride / write: pocos tokens claros (0 + 0 words)
• loud / lout: pocos tokens claros (0 + 0 words)

Grupos de referencia

Todavía hay menos de tres hablantes de tu origen — tu gráfico se compara con el grupo americano de lectura. Distancia entre tu sistema vocálico y el de cada grupo; menor es más cercano. El primer número es por posición en el gráfico; el segundo te sitúa por los rasgos medidos del acento (contrastes, posiciones, deslizamientos, la r tras vocal), que aún está en validación. Los grupos con menos de tres hablantes son una o dos personas, no un acento: lee esas filas como indicios.

1. General American (8 hablantes) — 0.23 por posición en el gráfico · 0.64 por rasgos
2. Irish English (Limerick) (4 hablantes) — 0.37 por posición en el gráfico · 0.77 por rasgos
3. Irish English (other regions) (3 hablantes) — 0.48 por posición en el gráfico · 0.88 por rasgos
4. Irish English (Dublin) (2 hablantes — provisional) — 0.62 por posición en el gráfico · 1.44 por rasgos
5. Other native English (British, Australian…) (5 hablantes) — 0.64 por posición en el gráfico · 1.28 por rasgos
6. Spanish L1 learners (2 hablantes — provisional) — 0.69 por posición en el gráfico · 1.16 por rasgos
7. Other-L1 learners (1 hablantes — provisional) — 0.75 por posición en el gráfico
8. Irish English (Ulster) (3 hablantes) — 0.8 por posición en el gráfico · 1.46 por rasgos

Your vowel space

Blue: the centre and spread of each of your vowel sets, in your own normalised space. Grey: where the native reference pool puts the same sets. Left is front, right is back, top is high. Show individual tokens

Each dot is one measured vowel token, coloured by set; hover for the word, its number (order of appearance in the passage), the part it came from and the raw formants. Hollow dots were excluded from the centroid as outliers. Token numbers match the per-set token list below.

What stands out

• FLEECE (the vowel of see) sits where the American reading pool puts it. Among the 8 other General American speakers here it sits with the group (0.15 units from their median position).
• KIT (the vowel of sit) sits where the American reading pool puts it. Among the 8 other General American speakers here it sits with the group (0.35 units from their median position).
• DRESS (the vowel of bed) sits where the American reading pool puts it. Among the 8 other General American speakers here it sits with the group (0.15 units from their median position).
• TRAP (the vowel of cat) sits where the American reading pool puts it. Among the 8 other General American speakers here it sits with the group (0.12 units from their median position).
• STRUT (the vowel of cup) sits where the American reading pool puts it. Among the 8 other General American speakers here it sits with the group (0.18 units from their median position).
• LOT (the vowel of odd) sits where the American reading pool puts it. Among the 8 other General American speakers here it sits with the group (0.34 units from their median position).
• THOUGHT (the vowel of dawn) is displaced 0.76 standard deviations from the American reading pool's position, 100% of the way toward LOT (the vowel of odd). Confidence interval 0.22–1.18 on 7 tokens. Among the 8 other General American speakers here it sits with the group (0.35 units from their median position).
• FOOT (the vowel of good) is displaced 1.1 standard deviations from the American reading pool's position, 69% of the way toward GOOSE (the vowel of food). Confidence interval 0.61–1.66 on 14 tokens. Among the 8 other General American speakers here it sits with the group (0.02 units from their median position).
• GOOSE (the vowel of food) sits where the American reading pool puts it. It is near the other General American speakers' position (0.57 units from their median).
• The KIT–FLEECE contrast is in the American reading pool's range (4.34; their median 2.59).
• The TRAP–STRUT contrast is larger than in 90% of the American reading pool (5.67 vs 2.66).
• FOOT–GOOSE separation 1.25 (no pool reference yet for this pair; a native speaker in this set measured 2.0–3.3).
• The DRESS–TRAP contrast is larger than in 90% of the American reading pool (2.57 vs 1.1).
• cot and caught: only 5 / 7 clean tokens — separation 1.04 is not distinguishable from shuffled labels (p = 0.298, chance median 0.77); undetermined.
• cot / caught: length cue runs the wrong way (coda voicing in the passage); not used for the verdict.
• cot / caught: the word list (controlled tokens, 6 + 5 words) says merged (separation 0.39, p = 0.819) where the passage says insufficient. The list is the better instrument for this pair; its verdict stands.
• put and putt: separation 2.82, p = 0.0 against shuffled labels — clearly distinct.
• The cat vowel before nasals (man, hand) is not raised — no pre-nasal split. (the 8 other General American speakers here: median 116 Hz, range 2–264; American median 116)
• pin and pen: distinct (F1 gap 74 Hz).
• GOOSE (food, two) F2 1385 Hz: fronted — the young-American / Southern-English pattern. (the 8 other General American speakers here: median -0.70 z, range -1.60–+0.10; American median -0.70)
• FACE (day, name) moves 11 Hz in F2 from a fifth to four-fifths of the way through the vowel — little movement at the midpoints (how much glide counts as a diphthong is being calibrated against the reference pools). (the 8 other General American speakers here: median 61 Hz, range 1–135; American median 61)
• GOAT (go, boat) moves 58 Hz in F2 from a fifth to four-fifths of the way through the vowel — little movement at the midpoints (how much glide counts as a diphthong is being calibrated against the reference pools). (the 8 other General American speakers here: median 142 Hz, range 35–266; American median 142)
• r after a vowel (nurse): F3 at the vowel's midpoint sits 910 Hz below your plain-vowel F3 on 9 tokens — rhotic (rhotic speakers drop 500–1100 Hz; non-rhotic 0–250). (the 8 other General American speakers here: median -836 Hz, range -979–-646; American median -836)
• aunt rhymes with ant.
• In free speech your contrasts KIT–FLEECE, TRAP–STRUT were smaller than when reading — normal: everyone's vowels move toward the centre in casual speech.

Each vowel set

Distance is in pooled within-set standard deviations; under 0.75 is "on target". "Shifted toward X 60%" means your centre has moved 60% of the way from the native position of this set to the native position of X.

Contrasts

Separation is the distance between two sets' clusters in your own space. Below the native 10th percentile means the pair overlaps more than 90% of native speakers' does.

Reading vs talking freely

Contrast separations while talking: KIT–FLEECE 1.92, FOOT–GOOSE 3.18, TRAP–STRUT 2.9, DRESS–TRAP 2.03, FOOT–STRUT 2.45, KIT–DRESS 2.1

The recording

Passage 37 s · 549 vowel tokens, 299 in content words, 246 used after quality control (15% dropped where the two aligners disagreed, 3% where the formant tracker was unsure). Reading check: mean coverage 0.982 — all parts read. Channel: Firefox, signal-to-noise 19.5 dB, usable bandwidth 11850 Hz — vowel-grade yes, fricative-grade yes.

Caveats

• This is exploratory research output, not an assessment. The measurements are automatic and some will be wrong.
• Vowel positions are medians over 5–25 tokens per set; sets with fewer than 8 tokens carry wide intervals and are marked.
• The grey reference is a pool of American read speech (LibriSpeech), named as such throughout. Where your own group has three or more other speakers, you are also compared with them; a group of three to eight is a small sample, so its range is shown, not a verdict.
• Nothing here is a judgment about individual words. Tokens the two aligners disagreed on, or the formant tracker was unsure of, were dropped before any number was computed.