Speech is a continuous flow that still requires discrete units to be understood

Evidence

Different languages vary in the form and number of basic units that are used to generate speech messages. Since there is temporal continuity of articulatory motion of the speech production mechanism, each unit is influenced by the preceding and following units. Therefore, speech is never composed of units in isolation but of a continuous flow of signal states. On the other hand, the structure of spoken language cannot be understood without some kind of discrete units that are concatenated into speech signals conveying linguistic information. This unit structure is reflected, more or less regularly, in written language. In languages such as English and French the written form (orthography) and phonetic transcription may often be quite different, while in languages at the other extreme, such as Finnish and Turkish, there is almost a one-to-one correspondence between the written and spoken forms. — Ville Pulkki, Matti Karjalainen, Communication Acoustics- An…, 565

Categorization

• DEFINE: Explains the foundational tension in how speech maps to discrete units, key to understanding acoustic speech coding.
• XREF: Connects to orthographic transparency debates in reading science and to the continuous-vs-discrete modeling choice in speech synthesis.
• corroborated: 3 independent passages support this claim — Ville Pulkki, Matti Karjalainen, Communication Acoustics- An…, 565