Liquid groups

[wehro]

All sources dealing with Latin hyphenation advise not to split liquid groups (except at word boundaries).
However, it is not completely clear, which groups of letters shall be treated as liquid groups.

Grammars of classical Latin use the notion of a muta cum liquida, i. e. a stop consonant (b, p, d, t, g, c) followed by a liquid consonant (l or r).
I have the impression that this definition is too general, because two possible combinations – dl and tl – never occur at the beginning of a native Latin word and nearly never within a native Latin word. The only exceptions I found are catlaster (syncopated from catulaster), catlitio (syncopated from catulitio), crustlum (syncopated from crustulum), mentla (syncopated from mentula), sitla (syncopated from situla), and titlus (syncopated from titulus) not taking into account several words of Greek origin.
The test list for liturgical Latin contains the hyphenations A-tlan-te-us, A-tlas, ca-tlas-tri, seu-tlo-pha-ce, Seu-tlu-sa, teu-tlo-pha-ce. This is consequent, but my feeling for language would prefer the hyphenation point between t and l (e. g. At-las), because tl is difficult to pronounce and I think it is not by accident that this group of consonants does not occur at the beginning of Latin words.
For dl the question is of academic nature because no words seem to be concerned.

Another difficult case are the groups fl and fr. These groups often occur at the beginning of Latin words, but f is no stop consonant; so fl and fr do not fulfil the definition of a muta cum liquida.
However, the Metodologia of Farina and Marinone, which Claudio used as a basis for the classical patterns and which I have cited here, advises to treat f the same way as a stop consonant in this respect. This is what the liturgical patterns do (in accordance with my personal feeling for language).
The only examples in the word list are aus-tro-a-fri-cus, Eu-fro-nius (which is however of Greek origin), and va-frum.
But a footnote in Allen's work Vox Latina (chapter 6: Quantity, p. 90) shows some doubts concerning these hyphenations:

The situation with f + liquid is not entirely clear. The grammarians equate f with Greek φ, though the classical φ was a plosive and not, like the Latin f, a fricative. And since f is the only non-plosive followed by a liquid in Latin, they simply state that a syllable has ‘common’ or ‘doubtful’ quantity before any consonant + liquid […]. In compounds, certainly, where both f and the liquid belong grammatically to the second element (e. g. re-fringo, re-fluo), the syllabic division may be, and usually is, re-fringo, etc., with consequently light first syllable. And when the grammarians seek to justify their statements, they invariably cite cases where the group belongs to a following word. Bede notes that such examples are invalid; but in fact […] even in the absence of word or morpheme boundaries light quantity is found before f + liquid, e. g. Rufras […] and Safroni […]; and in old senarii and scenic trochees not only in proper names (which are often subject to exceptional treatment) but also e. g. vafrae. The problem remains to what extent such treatment reflects presumed Greek parallels or actual Latin phonology.

The third problem concerning liquid groups are aspirated stops (ch, ph, th) followed by a liquid. I think these should be treated the same way as the unaspirated stops. So far, the test list for liturgical Latin is not consequent in one case: beth-le-hem (beside a-thle-tam and E-va-thlus). For this special case of thl, I would prefer the hyphenation point between th and l: Beth-le-hem, ath-le-tam, and E-vath-lus on the analogy of the case of unaspirated tl.