Why Can't I Understand Native English Speakers? 5 Reasons

Why Can't I Understand Native English Speakers?

Most English learners who cannot understand native speakers already know most of the words. The problem is finding those words in the sound stream. Native speakers join words together, drop and change sounds, and shrink small grammar words, so a familiar sentence stops sounding like its written words. Researchers call this a problem of decoding, or lexical segmentation.

John Field, a researcher in second language listening, made this argument in a 2003 paper in ELT Journal. Field called for "greater attention to the perceptual processes involved in second language listening" and proposed using basic auditory phonetics to classify, diagnose, and predict where learners fail to split speech into words. His point was that a listener who misses a sentence has often failed before meaning comes into it, at the stage of turning sound into words.

Later research put a number on the link. Wong and colleagues tested 64 Chinese undergraduate learners of English in a 2017 study in the Journal of Psycholinguistic Research. A dictation test of reduced forms in quiet conditions, covering processes such as assimilation, elision, and contraction, explained 28% of the variance in the students' general listening comprehension scores. That figure is correlational, but it shows that catching reduced, connected speech is a large share of what listening ability is.

The causes sort into five groups: connected speech, weak forms of small words, speaking speed, unfamiliar accents, and gaps in vocabulary. Each one has its own fix, covered below.

What Is Connected Speech, and Why Does It Hide English Words?

Connected speech is the way native English speakers join words in a phrase, changing the sounds at word edges so the mouth moves smoothly. Words link together (pick it up sounds like "pi-ki-tup"), sounds disappear (next day loses its /t/), and sounds change (did you becomes "didja"). The spaces between written words often do not exist in the sound.

Wong and colleagues' 2017 reduced forms test covered nine of these processes: assimilation, contraction, elision, flapping, glottalization, intrusion, juncture, palatalization, and vowel weakening. The four that learners meet most often are below.

The bad cat, do not, and could have examples come from Wong and colleagues' paper, which explains that native speakers follow an ease of articulation principle, so "connected speech consists highly of reduced forms, in which sounds are deleted or changed." The paper adds that reduced forms are hardest for learners whose first language does not reduce words in the same way.

Every rule, with audio, is in the Dobi Apps lesson on connected speech and how English words link together.

Why Do Small Words Like "To", "And", and "Of" Disappear in Fast English?

Small grammar words such as to, and, of, for, and can are almost always unstressed in natural English, so native speakers say them in a short, weak form: to becomes /tə/, and becomes /ən/, of can shrink to a single /ə/. A 2008 study by John Field found that learners recognize these function words much less reliably than content words.

Field's study, in TESOL Quarterly, tested intermediate-level second language listeners on which words they actually caught from connected speech. There were two plausible predictions. Function words are the most frequent words in English, so learners might catch them easily. Or function words are brief and carry little stress, so learners might miss them. The data supported the second.

Field reports that the gap held across learners with different first languages and at different levels of proficiency, so it is not a problem that only beginners have or that only speakers of one language face.

Missing a function word costs more than it seems. The small weak syllables carry tense, number, and the links between ideas: in "a cup of tea" the of shrinks to a single /ə/, and Siegel and Siegel's 2015 listening exercise contrasts "I want to go to the supermarket" with "I went to the supermarket", a difference of one vowel and one weak word. A listener who hears only the stressed content words gets a list of nouns and verbs and has to guess the grammar. English stress sets this up: content words take the stress, and the unstressed syllables around them shrink, a pattern explained in the Dobi Apps lesson on English word stress and rhythm. Learners who also find these sounds hard to produce can start with the hardest English sounds to pronounce and how to practice them.

How Fast Do Native English Speakers Actually Talk?

Native English conversation is fast. In the Switchboard corpus of 2,438 English telephone conversations, Jiahong Yuan, Mark Liberman, and Christopher Cieri measured an average of 196 words per minute across total conversation time, and 236 words per minute once silences and pauses were removed. Individual conversations ranged from 111 to 291 words per minute.

The same 2006 paper found that speaking rate in English conversation depends on context. Topic changed the average rate from 152 to nearly 170 words per minute in the Fisher corpus of conversations between strangers, and the authors report that "when talking with strangers or discussing certain topics, people tend to use longer turns but slower speech rates." Friends chatting about familiar things sit at the faster end.

Speed matters for learners because of what it removes. At higher rates, native speakers reduce more, and a listener has less time to match each chunk of sound to a word before the next chunk arrives. Yong Zhao's 1997 study in Applied Linguistics gave second language listeners control of the speech rate and found that "when given control, students' listening comprehension improved," and that the improvement came from slowing the rate down. How to use that finding with audiobooks, level by level, is in how audiobook playback speed affects English comprehension.

Why Do Accents and Unfamiliar Words Make Native English Harder?

An unfamiliar accent slows down word recognition even for native listeners, and unknown words leave gaps a listener cannot fill in real time. A 2009 study by Patti Adank and colleagues found that accent familiarity affected how fast listeners processed sentences in noise, and Paul Nation estimates that following unscripted speech comfortably takes 6,000 to 7,000 word families.

Adank, Evans, Stuart-Smith, and Scott, writing in the Journal of Experimental Psychology: Human Perception and Performance, played sentences to listeners at several levels of background noise in familiar native, unfamiliar native, and nonnative accents. Differences between native accents "influenced the speed of language processing under adverse listening conditions," and the cost of a nonnative accent was larger still. If native listeners pay a price for an unfamiliar accent, a learner trained mostly on one variety of English pays a bigger one when a speaker from another region starts talking.

Vocabulary is the second gap. Nation's 2006 corpus study in the Canadian Modern Language Review found that 3,000 word families plus proper nouns cover more than 95% of unscripted spoken English, and 6,000 to 7,000 reach 98%. Nation also warns that listening may need even higher coverage than reading, "to cope effectively with the transitory nature of spoken language": a listener cannot reread a sentence. The full set of figures, by text type and CEFR level, is in how many English words you need to read and speak.

Why Can I Read an English Sentence but Not Catch It When I Hear It?

Reading an English sentence shows where every word starts and ends, so the reader never has to segment the sound. Listening removes those spaces, and reduced word forms are hard to recognize without the words around them. A 2002 study by Mirjam Ernestus and colleagues found that even native listeners recognize highly reduced words well only in full context.

Ernestus, Baayen, and Schreuder ran two experiments on Dutch, published in Brain and Language, using words cut from casual speech. Listeners recognized highly reduced forms well only in full context, and in limited context their accuracy was strongly correlated with how reduced the word was. The study was in Dutch, but the mechanism is the same in English: a heavily reduced word is a smudge on its own and only becomes a word inside its sentence.

That explains a common learner experience. With subtitles, a clip is easy. Without them, the same clip turns into noise. The written text was doing the segmentation work, and the listener's ear has not learned to do it yet. The fix is to use text as a temporary support, not a permanent one: listen first, check the text, then listen again without it. Two posts go further on this: reading while listening to audiobooks and how captions and subtitles affect English learning.

What Practice Actually Helps You Understand Native English Speakers?

Practice that trains bottom-up decoding helps learners understand native English speakers. In a 2015 study by Joseph Siegel and Aki Siegel, Japanese university students who added short activities such as counting words, short transcriptions, and spotting connected speech to a normal course raised their dictation scores from 32.78 to 43.22, while a comparison group rose from 36.00 to 39.33.

The study, in Studies in Second Language Learning and Teaching, followed lower-intermediate learners through one semester. The treatment group spent about 10 minutes of some classes on six bottom-up activities, each used on three occasions. The treatment group also improved significantly on a general listening proficiency test while the comparison group did not. With 18 treatment and 26 comparison students analyzed, the authors say "we must exercise caution in finding any direct causal relationship due to the small number of participants." The learners were at roughly CEFR A2.

A learner can run the transcription activity alone: play one sentence, write it down, check it against the text, and note which missed words were weak forms or linked sounds, using material at the level described in comprehensible input and finding your English reading level.

Does Shadowing Help You Understand Native English Speakers?

Shadowing, repeating audio aloud almost at the same moment you hear it, trains the sound perception that native-speed listening depends on. Yo Hamada, a researcher at Akita University, explains that while shadowing, "the students' attention is mostly on catching the sounds, not meanings," which is exactly the bottom-up skill that ordinary comprehension exercises leave untrained.

Writing for TESL Ontario's CONTACT Magazine in 2018, Hamada summarizes his own 2016 research and gives practical rules for using shadowing as listening practice:

• Keep it short and intensive. Hamada reports a noticeable change from two 10 to 15 minute sessions a week for a month.
• Repeat each passage five or six times, then move on. The repetitions do not have to be consecutive.
• Understand the content first. Shadowing an unknown story is too demanding, so learners should know what the passage says before they shadow it.
• Shadow without the transcript. Reading while shadowing splits attention between sounds, letters, and meaning, and becomes a different task.

Hamada's article also separates two uses. If a learner cannot yet catch the sounds, shadowing should be treated as listening practice only. Once phoneme perception is in place, shadowing can move on to speaking and pronunciation work. The pronunciation side, and how to shadow with audiobooks, is covered in does shadowing audiobooks improve English pronunciation. Hamada also suggests speeding audio up to 1.5x for a round of shadowing, so the original speed then feels slower.

How Should You Practice Listening to Native-Speed English Each Week?

A practical weekly plan for understanding native-speed English mixes decoding work with ordinary listening for meaning. Use slowed audio and transcripts to find what you miss, dictation and shadowing to train your ear on it, and full-speed listening without text to test the result. Each method in the table below has research behind it.

A level-matched audiobook works for every row. Dobi Apps' Audiobook sorts its catalog by CEFR level and has an adjustable playback speed, so a learner can slow a chapter for dictation, shadow a passage, then replay it at full speed. The reason to give decoding this much time is Wong and colleagues' result: on its own, dictation of reduced forms in quiet conditions explained 28% of the variance in general listening comprehension.