This chapter examines the nature of tone sandhi and various other tonal mutations in Taiwanese Southern Min (TSM). Each base tone in this language corresponds to a specific sandhi tone, with sandhi resulting from two sets of tonal shifts: smooth tone chain shifts and checked tone chain shifts. Each shift modifies either register or pitch, but not both simultaneously.

Some experimental studies have reported low rates of sandhi application, suggesting limited productivity. However, evidence from both experiments using real words and corpus analysis reveals high rates of appropriate tonal alternations, indicating that productivity is the primary mechanism. Theoretical works have further elaborated the tonal alternations as systematic chain shifts, lending support to this productivity-based view. The evidence suggests that future models of TSM tone sandhi should primarily incorporate productive phonological processes, supplemented by selective lexical storage mechanisms for certain exceptional or high-frequency cases.

In diminutive suffixation, the tone of the pre-á syllable undergoes modification through dextrosinistral spreading of register and/or pitch from the -á suffix, whereby the derived [35] ([Lr, h][Hr, h]) tonal output emerges as a distinctive tone cluster. Conversely, in neutral tone operations, sinistrodextral tone spreading applies to a subsequent function word, which may alternatively acquire a low tone by default in the absence of such spreading.

This chapter examines the tone sandhi domains for pronouns, classifiers, and adverbs in TSM. Traditional X-bar theory projects pronouns under the noun phrase (NP). In this framework, a single pronoun forms a non-branching NP, which does not constitute a phonological phrase. Conversely, [Adj pronoun] constructions and coordinated pronouns form branching NPs, each establishing a phonological phrase. This approach, however, encounters a theoretical dilemma: a non-branching NP formed from a full noun constitutes a phonological phrase, while one formed from a pronoun does not.

The functional projection determiner phrase (DP) more accurately characterizes the phonological phrasing of pronouns: only a branching DP forms a phonological phrase, whereas an NP constitutes a phonological phrase regardless of branching status. In the analysis of classifiers, contemporary theories posit the classifier as the head of a classifier phrase (ClP), with the following noun as its complement, meaning [Num Cl] does not form an independent XP. This contrasts with the traditional view, which treats [Num Cl] as a modifier of the noun. In either case, [Num Cl] is not marked by a phonological phrase boundary. In contexts of nominal absence or topicalization, [Num Cl] may occupy the final position of the phonological phrase or undergo restructuring as a verbal adjunct if subject to semantic attenuation.

This chapter explores tone sandhi and tonal mutations in Taiwanese Southern Min (TSM). Each base tone corresponds to a specific sandhi tone, resulting from smooth or checked tone chain shifts that modify either register or pitch, but not both simultaneously. Despite some studies suggesting low productivity, real-word experiments and corpus analysis show high rates of tonal alternations, indicating that productivity is the primary mechanism. Future models of TSM tone sandhi should focus on productive phonological processes, with lexical storage for exceptional cases.

In diminutive suffixation, the pre-á syllable tone changes through register or pitch spreading from the -á suffix, forming a tone cluster. In neutral tone operations, spreading may lead to a default low tone.

Syllable contraction creates tone clusters from various tonal melodies, simplifying while preserving tonal information, typically through edge-in association and mora addition.

Trisyllabic reduplication involves an emphatic -á suffix on the leftmost syllable, with its high tone preserved as a floating tone if absent. Tetrasyllabic reduplication shows patterns of semantic emphasis. Some patterns form a single tone sandhi domain, while others split into two domains. The ABCC pattern, consisting of a subject NP and predicate VP, forms separate tonal domains.

Whereas Chapters 3–5 deal with different aspects of the representation of word meanings, Chapter 6 focuses on their organization. The two most well-studied domains of object concepts are animals and tools, and words for them (e.g., dog and scissors) appear to be organized as separate, category-specific circuits, each of which includes all three types of representation mentioned above – namely, sensory/motor representations in modal networks, integrated representations in the GSN/DMN, and purely verbal representations in the core language network. The contrasting specializations of these circuits for animal and tool concepts reflect many factors that involve both the learning environment and innate patterns of cortical connectivity. This chapter also shows that three other categories of lexically encoded object concepts similarly appear to have distinct neural substrates: plants (especially fruits and vegetables like apple and potato); people (including face parts like eye, body parts like hand, and unique individuals like Tom Hanks); and places (including generic scenes like beach and famous landmarks like the Eiffel Tower). What’s more, there are some hints that hundreds of other classes of word meanings may be distinguished, like in a thesaurus, by the fine-grained representational geometries of numerous cortical areas, especially those comprising the GSN/DMN.

The opening chapter provides a historical overview of Taiwanese Southern Min (TSM), tracing its development through the convergence of Zhangzhou and Quanzhou dialects. It introduces the subsequent chapters, each dedicated to specific phonological aspects: vowels, consonants, tones, syllable structure, segmental and tonal mutations, tonal domains, rhythm, and the evolving accent patterns of younger speakers, particularly the iGeneration Taiwanese Southern Min (iTSM), which represents a distinctive phonological profile.

The chapter also introduces the Taiwanese Romanization notation system alongside the International Phonetic Alphabet (IPA), the framework for data presentation throughout the study. Three robust TSM corpora, synthesized from earlier National Science Council research, provide the empirical foundation for the analysis. Statistical evaluations of the corpora support investigations into segmental transformations, tonal evolution, and prosodic patterns.

This introduction sets the stage for a comprehensive exploration of TSM phonology, encouraging readers to critically engage with the evidence and form independent interpretations. It prepares readers for a nuanced journey into the complexities of TSM phonology in the chapters ahead.

Folk verse occupies a middle ground between formal poetry and colloquial speech, maintaining poetic structure while preserving natural speech patterns. It adapts the metrical hierarchy of traditional verse but constructs feet based on metrical beats rather than syllables. This form distinguishes between masculine and feminine rhythms, with the former being predominant and the latter creating a softer tone. Function words and medial immediate constituents (ICs) tend to share a beat, and beat sharing serves as a strategy to achieve a masculine rhythm. However, a strong beat or a final beat cannot be shared, often resulting in an unparsed shared beat. An unparsed beat does not participate in clapping, and line-initial unparsed beats reflect extrametricality. A notable feature of folk verse is its abundant use of interjections, which enhance its rhythmic quality.

This chapter presents a detailed examination of TSM consonants, highlighting several key points. TSM’s consonant system displays partial asymmetry in voicing characteristics, and unlike in English, aspiration is a critical phonemic distinction.

Consonantal phonological features are systematically organized into distinct classificatory categories determined by their specification of natural segment classes. The hierarchical feature taxonomy encompasses four principal domains: major class features, which delineate fundamental segment types; laryngeal features, which characterize glottal states and phonation types; place features, which specify articulatory configurations and locations; and manner features, which define the type and degree of constriction in the vocal tract.

The alveolar lateral [l] in TSM serves a dual function: it integrates the alveolar /d/ within its articulation and exhibits free variation with [z]. Nasal stop onsets and oral voiced stop onsets occur in mutually exclusive environments, indicating complementary distribution. The glottal stop [ʔ] differs from the word-final codas [p, t, k] in that it is omitted when followed by another syllable and does not undergo gemination at the onset of a vowel-initial suffix, unlike [p, t, k]. Furthermore, [p, t, k] codas are underlyingly voiced, distinguishing them from nasal stop codas. This dichotomy between onset and coda stop consonants constitutes a salient feature of TSM’s phonological system.

This chapter explores various approaches to syllable division, the factors influencing syllabification, and the interaction of these principles with phonotactic constraints within a TSM language. Three prominent approaches to syllable constituency are discussed: Initial-Final, Onset-Rime, and Body-Coda. The Sonority Sequencing Principle (SSP) and Moraic Theory are identified as key factors governing syllabification in TSM. Notably, segment-mora mappings effectively account for checked syllables in TSM. Compensatory lengthening and closed syllable shortening phenomena provide additional evidence that voiceless stop codas in checked syllables are not extrasyllabic.

Significantly, these diverse approaches to syllable division are not mutually exclusive. A speaker’s linguistic intuition likely encompasses all these divisions, allowing them to leverage relevant phonotactic constraints at various levels, including Rime, Nucleus, Coda, Onset, Body, and even the entire syllable. When the domain is extended to encompass the entire syllable, the [nasal] feature is prohibited from spreading across the boundaries of Body and Coda. Consequently, nasality in this language manifests as a crisp edge phenomenon.

This chapter examines segmental mutations and processes in Taiwanese Southern Min (TSM), including assimilation, dissimilation, gemination, syllabic reduction, contraction, and vernacular-literary phonological differences.

Assimilation occurs in place (e.g., homorganic nasal assimilation, optional palatalization) and manner (e.g., nasal harmony). Dissimilation prevents adjacent identical [back] features in diphthongs and avoids co-occurrence of [labial] and [dorsal] features in onset glides and coda consonants.

Gemination lengthens glides, nasals, and stops. Stop codas of checked syllables are voiced and undergo gemination when followed by a vowel-initial function word; otherwise, they undergo final devoicing. Gemination is constrained to the prosodic word domain, and the chapter suggests a gradient interpretation of geminate inalterability in some systems.

Syllable reduction leads to contraction, with nuclear segments potentially undergoing transformations like devocalization, merger, or nasalization. Contraction follows an edge-in paradigm, with sonority-based priority.

Vernacular-literary differences include onset consonants in the literary register undergoing pharyngealization and velarization, and vowels alternating through rounding and derounding, reflecting a trend toward unmarked phonological structures.

The phonological phrasing of adverbs reveals a complex interplay of syntactic positioning, semantic modification scope, and semantic interpretive mode. Several nuanced patterns emerge from this linguistic phenomenon. In syntactic medial position, when an adverb operates at the sentence scope, a phonological phrase boundary materializes subsequently; however, this boundary is precluded following a predicate-scope adverb. In syntactic initial position, a phonological phrase boundary manifests after a speaker-oriented adverb with a subjective interpretative mode, yet remains absent in instances of adverbs with an objective interpretative stance. Moreover, such a boundary occurs after a focused polymorphic-scope adverb, while being systematically absent following a focused predicate-scope adverb. In essence, the focal adverbial tonal domain is fundamentally circumscribed by its scope of semantic modification, demonstrating the complex relationship between syntactic positioning and semantic interpretation.

Function words typically exhibit extraprosodicity at the phonological phrase level and are consequently exempt from tone sandhi processes. These elements, however, are reintegrated at the intonational phrase level, where they become subject to neutral tone operations. When function words receive focus, they no longer exhibit extraprosodicity and are thus subject to tone sandhi processes.

Finally, Chapter 8 rounds off the book by summarizing the main conclusions, highlighting several directions for future research, and considering the broader implications of the neuroscience of word meaning.

Through syllable contraction, sandhi, base, and neutral tonal melodies merge, resulting in single tones or tone clusters. Simplification follows via edge-in association, which preserves the edge tones. In cases where clusters are longer, mora addition is employed.

Trisyllabic reduplication involves an emphatic -á suffix attached to the leftmost syllable. When this emphatic -á is absent in triplication, its high tone persists as a floating tone. The initial syllable of triplication, functioning as the prosodic head and favoring a high tone, bears focal stress. If this syllable surfaces with a low-register sandhi tone, it recruits the floating high tone to form a tone cluster. However, if it already carries a high-register tone and satisfies prosodic prominence, the floating high tone remains redundant.

Tetrasyllabic reduplication exhibits distinct phonological and semantic patterns in its various manifestations. The ABAB configuration functions to attenuate semantic meaning, whereas AABB, ABAC, and ACBC patterns serve to intensify semantic content. From a prosodic perspective, ABAB configurations consistently operate as unified tone sandhi domains. In contrast, AABB, ABAC, and ACBC patterns demonstrate prosodic flexibility, potentially functioning either as single unified domains or bifurcating into two discrete domains. It is critical to note that the ABCC pattern, despite its superficial similarity, does not represent genuine tetrasyllabic reduplication; instead, it comprises two distinct phrasal elements that invariably constitute separate tonal domains.