Broad Phonetic Transcription: IPA Basics and Practice

Connected speech operates through five distinct forces that explain why native speakers sound different from textbook English. Reduction is the most widespread phenomenon: content words (nouns, verbs, adjectives, adverbs) are pronounced fully, while function words (articles, prepositions, pronouns, auxiliary verbs, conjunctions) are reduced, weakened, or almost swallowed. Examples include 'to' becoming 'tuh', 'of' becoming 'uv', 'and' becoming 'n', 'should have' becoming 'shoulda'. Elision is when sounds disappear completely: T-elision (important becomes 'imporant'), D-elision (old man becomes 'old man'), H-dropping (he becomes 'e'), and consonant cluster reduction. Assimilation is when sounds influence each other at word boundaries: T+Y becomes CH (did you), D+Y becomes J (did your), S+Y becomes SH (this year), N+Y becomes NY (in your). Linking creates smooth sound streams: consonant-to-vowel linking (not at all), vowel-to-vowel linking with intrusive sounds (go away becomes 'go w away'), and consonant-to-consonant linking (black coffee).

Connected speech processes are phonological phenomena that occur when sounds interact at word boundaries during rapid, casual speech, including assimilation (where sounds become more similar to neighboring sounds through changes in place, manner, or voicing), coalescence (where two consonants merge into one affricate), elision (where sounds disappear to avoid complex clusters), linking R (where R is pronounced before vowel-initial words), intrusive R (where R is inserted between non-closed vowels), and juncture (the relationship between sounds at word boundaries affecting their pronunciation).

This section examines processes in connected speech. Assimilation occurs when adjacent sounds influence each other, such as 'I've got to go' where 've' assimilates. Nasal assimilation causes preceding sounds to become nasalized before nasal consonants. Elision involves dropping sounds that might be present, like the 'd' in 'and' when surrounded by nasals. These processes maximize meaning and minimize effort, occurring in everyone's normal speech. They should not be regarded as sloppiness but as efficiency mechanisms that speakers subconsciously employ.

Connected speech refers to how words combine in actual speech, causing segments to weaken, drop out, or change. English speech is organized into tone units with few stressed words, triggering phonological processes. Assimilation occurs when adjacent phonemes influence each other to become more alike, either through voicing changes or place of articulation. Regressive assimilation happens when word-final consonants adopt characteristics of the following word's initial consonant, as in '10 coins' becoming '10 coins'. Progressive assimilation works in the opposite direction, where preceding sounds influence subsequent ones, as seen in 'ribbon' becoming 'ribbon'.

Connected Speech is the phenomenon where native English speakers link words together for smoother, faster communication. Three key processes include: Assimilation, where sounds change when words connect (e.g., 'don't you' becomes 'don't youa' with 'd' changing to 'y' sound); Elision, where sounds are omitted (e.g., 'next door' drops the 't', 'must go' drops the 't'); and Linking, where consonants connect to following vowels (e.g., 'this afternoon' links 's' to 'a'). These processes create natural-sounding English by reducing distinct word boundaries.

Acoustic phonetics studies the physical properties of speech sounds. A spectrogram is a visual representation of sound produced by software like Praat, displaying patterns of speech. Voice sounds appear as vertical striations indicating vocal fold vibration. Different consonant types show distinct patterns: bilabial sounds (lips) have low F2 and F3; alveolar sounds (tongue against alveolar ridge) show F2 at 1700-1800 Hz; velar sounds (back of mouth) show high F2; retroflex sounds show downward bends in F3 and F4. Stop sounds show gaps and bursts; fricatives show noisy high-frequency patterns; nasals have vowel-like formants; approximants have rapidly changing formants. Spectrograms are valuable for analyzing speech sounds, pitch, intonation, and are used in speech recognition and machine translation.

This tutorial demonstrates how to use Praat software for acoustic speech analysis, covering audio recording and import, text grid creation with three tiers (orthography, phonetic transcription, and realization), and spectrogram analysis including voicing amplitude, pitch measurement (165-255 Hz for women), and formant analysis (F1 and F2) to characterize vowel sounds and identify consonants through spectrogram patterns.

Acoustic phonetics studies speech sounds through physical properties like frequency, intensity, and duration using graphical analysis. The source-filter theory divides speech production into the source (vocal fold vibration creating pitch) and filter (vocal tract shaping sound into phonemes). The perturbation theory explains vowel formation through vocal tract constrictions affecting resonance frequencies. Praat software enables practical analysis through its interface with Praat Objects for file management, View and Edit for oscillogram and spectrogram visualization, and spectrogram settings (wide band for temporal resolution, narrow band for frequency detail).

Praat is a free, open-source software developed by University of Amsterdam professors for acoustic analysis, widely used in academia because it is powerful, lightweight, and allows script creation for automation. Users can create synthetic tones with specific frequencies - 440 Hz (A4), 220 Hz (A3, one octave below), and 880 Hz (A5, one octave above). Praat displays two main visualizations: the oscillogram (top window) shows the waveform of sound, displaying amplitude over time, while the spectrogram (bottom window) shows frequency content over time with darker areas indicating stronger frequencies. Pure tones produce simple sine waves, but human voice produces complex waves that are combinations of multiple simple waves at different frequencies. The source-filter model explains speech production: vocal cords (source) generate basic sound, while articulators above the glottis (filter) modify this sound to produce different vowels and consonants.

This advanced phonetics and phonology course requires prior completion of introductory linguistics, where students should have learned articulators, place and manner of articulation, and voicing. Phonetics is a hard-core science involving vocal tract mechanics, air stream mechanisms, and acoustic analysis through waveform examination. Students will learn to use software like Praat to record and analyze their own speech. The course covers articulatory phonetics (how sounds are produced) and acoustic phonetics (how sounds travel through the air). Students will learn IPA symbols beyond English sounds, including diacritics for nasalization and unreleased consonants. The course distinguishes phonetics (physical sound production) from phonology (abstract sound patterns and rules).

This section explores the practical applications and professional fields where phonetics and phonology knowledge is essential. Historically, these areas received limited attention in education, but this is changing as students recognize that understanding language requires phonological knowledge alongside morphological and syntactic knowledge. Technology has advanced with tools like Praat software for acoustic analysis. Applications include literacy education (teachers need phonetic knowledge to address student needs and linguistic prejudice), language teaching (comparing sound systems of native and foreign languages), and speech-language pathology (understanding articulatory and cognitive mechanisms for speech disorders). Forensic phonetics has grown significantly, using laboratory instruments to identify speakers by comparing voice samples. Speech technology works on synthesis (computers vocalizing text), recognition (machines understanding speech), and interaction (human-machine dialogue). Translation professionals need knowledge of sound systems in both source and target languages. Understanding phonetics and phonology is essential across multiple professional fields.

Phonetics has numerous practical applications across different fields: (1) Language documentation - studying how sounds differ across languages, such as sounds present in English but not Arabic or French; (2) Speech pathology - helping people with speech production problems; (3) Speech technology - developing automatic speech recognition systems like Google Assistant, Cortana, and Siri; (4) Pronunciation teaching - teaching correct pronunciation techniques; (5) Cognitive science - studying the brain processes involved in speaking and listening; (6) Acting and dialect coaching - helping actors speak different dialects convincingly.

Speech-language therapists must master phonetic transcription using the International Phonetic Alphabet (IPA) because phonetics is the scientific study of speech sounds that enables accurate pronunciation across different languages; unlike musical notation where musicians read symbols rather than note names, therapists should transcribe speech using phonetic symbols rather than standard orthography to professionally document articulation patterns and accurately represent how sounds are produced, which is essential for effective language therapy and foreign language pronunciation guidance.

Phonetics and phonology serve specialized professional fields: (1) Forensic linguistics uses voice analysis to identify speakers in legal contexts, with regional accents revealing geographic origins; (2) Creating alphabets for unwritten languages requires identifying phonemes versus allophones through fieldwork and careful discrimination between similar sounds. Both applications demonstrate how systematic phonetic analysis provides objective tools for solving practical problems in law enforcement and language preservation.

Phonetics is important in audio forensics for legal purposes—phoneticians analyze pronunciation to determine speaker origin. Phonetic transcription is used in court reports. Automatic transcribers (like Plus Ultra from USP) are fast but based on specific dialects (São Paulo), potentially missing details like affricates. Manual verification is important, and dialect should prevail in transcription.