Their time reference would be the same as that of our ancestors who lived without clocks.

They lived by the sunrise and sunset, by the tides, by the changes in the moon, by the changing seasons and their own internal body clock.

They would have a concept of days, months and years by carefully observing these events.

There is an interesting article on the subject here:

http://www.cosmosmagazine.com/features/print/2187/time-warp

Particularly the piece about Michel Siffre (a bit like the word chiffre, which means numeral, as on a clock - hmmm).

From my view, the lack of language to form a view of time seriously impedes its perceptibility. For someone raised completely outside of a context where there's simply an inherited understanding of time, there's likely nothing obvious that links memories to the past. So, for that individual without the language for it, the idea that things happen in a particular order is a very advanced concept and not necessarily intuitive. In fact, it's not intuitive-- Einstein wrestled with this. For that individual without the construct of time, things are simply happening now and now and now, and the brain makes guesses about what will happen next based on what it perceives in the present. The conversation for time doesn't weigh into it.

I posit that the perception of time's passing (whether the experience of the observer's passage through time is faster or slower than expectations) relies very heavily on the observer having a view that, first, there is such a thing as time AND there are events expected to happen in the future AND the viewer has enough experience with time itself to evaluate whether he/she is moving toward the expected event at the usual rate.

For a child raised without a concept of time or any of the language that's time-based, nor any experience of measuring time, it's likely that time does not exist for them and that they likely experience everything as the present-- they don't have an awareness of "time passing" (a highly intellectual concept).

A related question is how different linguistic models of time (v. past/present/future or seconds/minutes/hours) affect the observer's experience of time-- which I'm now curious about.

Why do we perceive time to flow faster as we get older?  Would entropy exist without consciousness?

It depends on the range of conditional variables the child is proximally exposed to, and the genetically inherited range.

A 'normal' child would adapt less favorably to deficient stimuli that has factored more into their lineage's momentum. In other words, take away exposure to sunlight differentials, without applying a few substitutes to offset the deficiency, and the initial adaptability set of the birth-stage, skews to deficiency. If, however, you have a child that enjoys a more adaptable lineage - even genes that have skipped a few generations, then that child will adapt better - barring, of course, severe variables that keep the child at a limited math.

It's interesting to note that many behavioral scientists now believe that we have progressed to a state that many early deficiencies - including genetic predispositions, are more malleable as we accelerate technological variables (which, of course, feedback to generational factors in evolutionary scale).

A study in Technological Singularity might add value.