Marijn van ‘t Veer – Introduction to Element Theory (two weeks)


Course Description. In this course, we will explore Element Theory, an alternative framework of (sub-)segmental representation. Element Theory differs from feature theory in three major ways:

  • Elements are defined as acoustic targets, rather than articulatory events
  • Autonomous Interpretation: every element can be phonetically interpreted on its own terms. Hence, a segment can be simple (consist of a single element) or complex (multiple elements)
  • Vowel-Consonant uniformity:  the same set of elements can be interpreted in both nuclear (vowel) and non-nuclear (consonantal) positions

These three aspects of Element Theory are to a large extent connected,  and this connectedness will be a recurrent theme throughout the course. For example, the autonomous interpretation of elements follows (in part) from their identity as acoustic targets (and hence their agnosticism with respect to articulatory gestures). The element |I| is defined as a spectral pattern with intensity peaks at high and   low frequency bands, and a dip in energy in between (low F1, high F2).  This is fully realised in the vowel [i], and hence this vowel is represented simply by |I|. In addition, it is also apparent in the acoustic signature of [e], and therefore it must follow that [e] also has |I| in its structural make up. Furthermore, the same spectral pattern is realised in the glide [j], and so we can see that [i] and [j] are segmentally identical, the difference stemming from a different prosodic position (nuclear vs. non-nuclear).

We will work with Backley’s set of six elements, and see how they combine to represent any segment inventory. Furthermore, we will discuss syllable structure and its interactions with elements, and how elements function in autosegmental ways in phonological processes. We shall also pay attention to the role of contrast in Element Theory, which is arguably different from how it functions in most Feature Theories. The way in which phonological research hypotheses are constructed, predictions are derived and tested, and arguments are evaluated forms a major part of each lecture.

Class descriptions

Class 1: Introduction. In this class, we will foreshadow the weeks to come, have a first look at the book and start with the basic concepts/terminology. We will begin our introduction to Element Theory, paying specific attention to three main aspects: Acoustic targets, Autonomous Interpretation and Vowel-Consonant Uniformity.

Reading Material: Backley, chapter 1

You’re invited to introduce yourself in class: where are you from, what is your native language, other languages you speak, what is your background in phonology and/or phonetics, what are your interests, what do you hope to take away from this course, do you have any special requests.

Class 2:  Elements for Vowels.  The resonance elements |I|, |A| and |U| will be inspected closely, along with their phonetic interpretations as the corner vowels [a, i, u]. We will discuss “empty” vowels, before continuing to complex vowels, using color theory as a metaphor. Next, we will continue our discussion of complex vowels. We will introduce the concept of Headedness, and see how the resonance elements differ in what they contribute to complex vowels. We will discuss both synchronic and diachronic examples of complex vowels, the acoustic correlates of each element and the natural classes denoted by each.

Reading Material: Backley, chapter 2

Propose a representation ofthe vowel inventory of your native language using the three resonance elements.

Class 3: Place Elements for Consonants. During the first part of today’s class we will discuss your homework. One way in which Element theory differs from feature theories is that there is no independent Place of Articulation dimension. We will illustrate how the IPA consonant chart is inherently articulatory organised, and what an acoustic organisation might look like. We begin our exploration of consonants with glides, which are straightforwardly [j, w] in the case of |I| and |U| respectively, whereas the glide-vowel relationship for |A| requires more attention. Analogous to our discussion of the three resonance elements in the context of vowels, we will discuss each in turn, focusing on acoustic characteristics and phonological evidence for the natural class each defines.

Reading Material: Backley, chapter 3

Class 4: Manner elements in consonants, pt. I. Just as we have three resonance elements, there are three non-resonance elements: |L|, |H|, |P|. The first two specify a low and high spectral center of gravity respectively, whereas the third specifies a sudden drop in energy. The |L| element is present in voiced obstruents (we will discuss the merits of so-called laryngeal realism in this class) and nasals. Voicing, that is to say, the presence of F0 in the spectrum, pulls down the center of gravity towards the lower region. |H|, on the other hand, aims for a high-frequency concentration of energy. This is typically realised as frication, or even stridency. The sudden drop in energy typical of stops, finally, is specified by |P|, the stop- element. Again, we will examine the three new elements from an acoustic and phonological point of view.

Reading Material: Backley, chapter 4

Class 5: Manner elements in consonants, pt. II.  The non-resonance elements introduce a rather different classification of inventories, and in this session, we will focus further on issues where natural classes appear to be different than in feature theory. For example, whereas in feature theory obstruents are voiced, and sonorant voicing is a matter of debate, the class of sonorants is not even positively defined in element theory. Instead, voiced obstruents and nasals share |L|, whereas liquids and glides do not (in fact, they lack non-resonance elements altogether). We will examine the predictions about natural classes in detail, as we did with the resonance elements earlier, focusing on phenomena such as nasal harmony, lenition and laryngeal realism.

Reading Material: Backley, chapter 4

Propose a representation of the consonant inventory of your native language using all  six elements

Class 6: Liquids. Liquids are notoriously difficult to represent in feature the- ory, having a broad range of phonetic diversity versus a quite well-defined phono- tactic distribution. We will begin our exploration today with discussing whether there is such a thing as a ‘class of liquids’ in the first place, before considering the way liquids are represented in element theory: lacking a non-resonance element, rhotics are on a par with glides and laterals are complex glides.  Furthermore, we will discuss ongoing research on the class-hood of liquids and glides.

Reading Material: Backley, chapter 5. De Kok et al, 2018

Class 7: Elements in Syllables. A recurring theme in this course has been that in element theory, phonological behaviour is the prime motivator for assigning sub segmental structure. That is to say, if segment X undergoes the same set of rules or is subject to the same distributional restrictions as segment Y, they must form a natural class and hence share an element. In today’s class, we will apply the same logic to syllable structure. If a certain syllabic position A is host to the same subset of segments, or subject to the same phonological alternations as position B, they must be of the same type and vice versa. We will focus specifically on the purported uniformity of codas and word-final consonants. While the syllable as a level of organisation was introduced in generative phonology to account for the apparent overlap of rules specifying either C or #, this overlap turns out to be not exact enough to warrant coda status for word-final consonants, or some word-internal consonant clusters. As we used class behaviour to determine phonological identity in earlier classes, we will do so again with regard to supra-segmental structure. We will see that the distributional properties of word-final consonants, while resembling onsets more than codas, are not or not always be identical to the former. This observation allows us to begin our exploration of Licensing, which will be the topic of the next class. We will discuss, furthermore, a striking example of liquid-glide uniformity in child language, where the learners sacrifice segmental faithfulness in order to express prosodic asymmetry.

Reading Material: Harris and Gussmann, 2002. Inkelas and Rose, 2008

Class 8: Licensing Inheritance. We now have a fairly complete picture of sub-segmental and supra-segmental structure, so in this class we will explore how segments/elements relate to each other in different positions. In today’s class, we will discuss the two types of neutralisation identified by Harris: reductive and as- similative. In both cases, the number of contrasts that can be expressed in a given position is smaller than in another position (in the same language), the difference being whether the segmental context plays a role in determining the outcome. A typical example of reductive neutralisation is Final Obstruent Devoicing, whereas a typical example of assimilative neutralisation is nasal place assimilation. So, whereas the two types of neutralisation are segmentally different, Harris proposes that positionally, they are identical and hence should be unified.  The solution lies in the asymmetry between strong and weak positions. The difference between strong and weak positions can be derived from three licensing principals: general licensing, onset licensing and coda licensing. Harris’ innovation is that the seg- mental licensing potential (a-licensing) of any position is inherited from its own licensor (p-licensing). The prediction is that the longer the licensing path from any position to its head, the weaker its licensing potential. In other words, the more “peripheral” a position, the smaller the number of contrasts it can express. A main secondary topic in today’s class is how licensing relates to association lines in autosegmental phonology. The by now well familiar principle of Autonomous In- terpretation has been demonstrated to make the distinction between two levels of phonological representation obsolete in earlier classes, and now we can see how the non-derivational tendencies of autosegmental phonology can be applied rigorously through the concept of licensing.

Reading Material: Harris, 1997

Class 9: Sonorants and Lenition. We have seen how concepts underpin- ning Element Theory and Licensing can account for a range of phenomena. One such phenomenon that we discussed at multiple times is lenition. Because elements are autonomous, Element Theory is uniquely capable of expressing the relation between positional and segmental strength: weak positions are further away from head positions in terms of licensing, and hence can accommodate less phonological complexity (i.e., less elements).

A widespread phenomenon that has many of the characteristics of lenition, but has so far eluded a satisfactory analysis, is Intervocalic Voicing. Targeting weak positions, often accompanied by spirantisation (another lenition process), and by the looks of it assimilatory, analyses in terms of spreading (be it of the feature [Voice] or the element |L| run into multiple problems. Starting from a typological observation about the frequent occurrence of unpaired non-sibilant voiced fricatives, we argue that diachronically, these are the result of Intervocalic Voicing, and that Intervocalic Voicing in fact is not voicing at all, but rather the loss of non-resonant elements in weak position, which we dub reduction to sonorancy. This hypothesis is backed by a broad typological survey, in which we  how that  in  all but  two  of the cases where evidence is available, the unpaired non-sibilant voiced fricatives form display natural class behaviour (synchronically) with sonorants rather than obstruents.

Reading Material: Botma and van ’t Veer, 2013

Please think of at least three questions you would like to have answered in the final session

Class 10: Loose ends, dotting i’s, crossing t’s. You get to choose today’s topic(s), in part by submitting yesterday’s homework, possibly in part by submitting special requests for topics that have not yet been addressed (NB: the earlier you submit your requests, the bigger the chance that I will be able to come up with a somewhat satisfying response).


Backley, P. (2011). An introduction to element theory. Edinburgh: Edinburgh University Press.

Botma, B., & van ’t Veer, B. M. (2013). A fraction too much friction: The phonological status of voiced fricatives. In S. Aalberse & A. Auer (Eds.), Linguistics in the Netherlands 2013 (p. xx-xx). Amsterdam: John Benjamins Publishing Company.

Harris, J. (1997). Licensing inheritance: an integrated theory of neutralisation. Phonology, 14 , 315-370.

Harris, J., & Gussmann, E. (2002). Word-final onsets. UCL Working Papers in Linguistics.

Inkelas, S., & Rose, Y. (2008). Positional neutralization: A case study from child language. Language, 83 .

Kenneth  de  Kok,  B.  B.,  &  van  ’t  Veer,  M.  (2018).   Glides  and  laryngeals  as a structural class. In B. L. Bruyn & J. Berns (Eds.), Linguistics in the netherlands (Vol. 35). Amsterdam: John Benjamins Publishing Company.